Information Structure and the Dynamics of Language Acquisition [1 ed.] 9789027296436, 9789027241375

Citation preview

Information Structure and the Dynamics of Language Acquisition

Studies in Bilingualism (SiBil)

Editors Kees de Bot University of Groningen

Thom Huebner San José State University

Editorial Board Michael Clyne, University of Melbourne Kathryn Davis, University of Hawaii at Manoa Joshua Fishman, Yeshiva University François Grosjean, Université de Neuchâtel Wolfgang Klein, Max Planck Institut für Psycholinguistik Georges Lüdi, University of Basel Christina Bratt Paulston, University of Pittsburgh Suzanne Romaine, Merton College, Oxford Merrill Swain, Ontario Institute for Studies in Education Richard Tucker, Carnegie Mellon University

Volume 26 Information Structure and the Dynamics of Language Acquisition Edited by Christine Dimroth and Marianne Starren

Information Structure and the Dynamics of Language Acquisition Edited by

Christine Dimroth Max Planck Institute, Nijmegen

Marianne Starren University of Nijmegen

John Benjamins Publishing Company Amsterdam
/
Philadelphia

8

TM

The paper used in this publication meets the minimum requirements of American National Standard for Information Sciences – Permanence of Paper for Printed Library Materials, ansi z39.48-1984.

Library of Congress Cataloging-in-Publication Data Information structure and the dynamics of language acquisition / edited by Christine Dimroth and Marianne Starren. p. cm. (Studies in Bilingualism, issn 0928–1533 ; v. 26) Includes bibliographical references and index. 1. Language acquisition. 2. Scope (Linguistics) 3. Anaphora (Linguistics) I. Dimroth, Christine. II. Starren, Marianne. III. Series. P118.I436 2003 401’.93-dc21 isbn 90 272 4137 6 (Eur.) / 1 58811 412 0 (US) (Hb; alk. paper)

2003052268

© 2003 – John Benjamins B.V. No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publisher. John Benjamins Publishing Co. · P.O. Box 36224 · 1020 me Amsterdam · The Netherlands John Benjamins North America · P.O. Box 27519 · Philadelphia pa 19118-0519 · usa

Table of contents

Introduction

1

I. Finiteness and scope relations Development of verb morphology and finiteness in children and adults acquiring French Suzanne Schlyter “Tinkering” with chunks: Form-oriented strategies and idiosyncratic utterance patterns without functional implications in the IL of Turkish speaking children learning German Stefanie Haberzettl Finiteness in Germanic languages: A stage-model for first and second language development Christine Dimroth, Petra Gretsch, Peter Jordens, Clive Perdue, and Marianne Starren

15

45

65

On the similarities of L1 and L2 acquisition: How German children anchor utterances in time Petra Gretsch

95

Negation and relational predicates in French and English as second languages Patrizia Giuliano

119

The copula in learner Italian: Finiteness and verbal inflection Giuliano Bernini The interaction between the development of verb morphology and the acquisition of temporal adverbs of contrast: A longitudinal study in French, English and German L2 Sandra Benazzo

159

187



Table of contents

Merging scope particles. Word order variation and the acquisition of aussi and ook in a bilingual context Aafke Hulk Creole prototypes as basic varieties and inflectional morphology Angelika Becker and Tonjes Veenstra

211 235

II. Anaphoric relations Information structure in narratives and the role of grammaticised knowledge: A study of adult French and German learners of English Mary Carroll and Monique Lambert

267

Acquiring the linkage between syntactic, semantic and informational roles in narratives by Spanish learners of German Jorge Murcia-Serra

289

Gestures, referents, and anaphoric linkage in learner varieties Marianne Gullberg

311

The development of anaphoric means to refer to space and entities in the acquisition of French by Polish learners Marzena Watorek

329

Subject index

357

Introduction

The concept of information structure has occurred in a wide range of linguistic approaches, variously associated with pragmatics, (formal) semantics, discourse analyses, or with a combination of these (Lambrecht 1994; Rooth 1985; Vallduví 1992). The papers in this volume focus on the impact of information structure on the process of language acquisition thereby taking also a wide range of linguistic approaches into account. They start from an empirical point of view, and examine data from natural first and second language acquisition. The acquisition data which are investigated in the present volume cover a wide range of varieties, from early learner language to native speaker production and from gesture to Creole prototypes. However, the underlying themes of the contributions show great coherence. The central theme is the impact of the interplay between principles of information structure and linguistic structure on the functioning and development of the learner’s system. The structure of very basic learner utterances as well as the construction of coherent discourse in very advanced learners depend crucially on the linguistic repertoire available at a given stage. The papers in this volume investigate how the need to express more complicated information structures pushes the acquisition process. They show how principles of information organization can have a different impact on the shape of learner utterances at different developmental stages of language acquisition. Information structure is thus considered as an important factor for the dynamics of language acquisition. This is especially important for natural (non-guided) language acquisition, where learners have to anchor their utterances in context in order to construct a coherent body of information. In very early stages of acquisition, learner utterances typically lack such specifications of time, place, or protagonist. In the following example from Schlyter (this volume), anchoring in the context is provided by the target speaker (mother) by morpho-syntactic coding of finiteness on the verb. (1) Mother: Child:

quand est-ce que tu mets ton chapeau, Jean? when do you put your hat on, Jean? courir run-inf



Introduction

Mother:

Courir. Quand on va courir. run. when we will run

A lot of earlier research on finiteness is mainly concerned with the acquisition of the linguistic marking of finiteness which occurs only in later stages of L1 and L2 acquisition. The papers in this volume tackle the question whether the semantic concept of finiteness (anchoring to time, space, protagonist) shapes the form of utterances before the target language means of finiteness marking occur, or if more formal mechanisms are responsible for the emergence of morpho-syntactic markings. Examples can again be found at the beginning of the acquisition process, where shared principles of information organization shape the linerarization of linguistic elements in an utterance. Word order seems to follow a basic rule: learners first refer to the topic situation (protagonist, time, space) the utterance is about and then express some state of affairs that holds for this topic. Compare the following examples from Gretsch (2) and Dimroth et al. (3): (2) child learner:

ganze hase

(3) adult learner: vandaag

kaputt hoofdpijn

These examples show that a lot of information is underspecified which makes the interpretation of these basic utterances very context dependent. Without knowing which situation is actually talked about it is impossible to infer if (2) refers to the present or to the past, or to understand who had a headache on the day when (3) was uttered. We can see that these utterances need a lot of additional information and scaffolding from the linguistic or extralinguistic context in order to be interpretable. It can be assumed that the communicative need to become a more independent speaker and the need to talk about more complex situations pushes learners to acquire new linguistic means. Both these needs crucially involve the capacity of constructing pieces of discourse instead of isolated utterances. In order to do so, learners have to adapt the form of their utterances to the intended macro-structure of information organization. Such a more complex information structure might for example involve the integration of scope sensitive operators like negation or scope particles (i.e. elements like also, only etc.). The L1 and L2 learners in the following examples from Hulk (4) and Giuliano (5) try to express that only the underlined part of the utterances are in the scope of the respective operators. (4) Mother: Child:

il est cassé celui-là he is broken that one aussi cassé also broken

Introduction

Mother:

pas aussi cassé, Anouk, il est cassé aussi not also broken, A., he is broken also

(5) Native Speaker: It (= the use of fireworks in Italy) is against the law? Adult learner: no everything *contro la legge* not all of them are against the law The form of these utterances is adapted to properties of the linguistic and/or extralinguistic context. The child in (4) marks that a piece of information is added to some already established set of broken objects. The adult L2 learner in (5) restricts the validity of an earlier claim (fireworks being against the law) by saying that this claim does not apply to all entities that fall under the header of fireworks. In order to do so, learners have to indicate precisely which part of their utterances are affected by negation or scope particles: They have to develop linguistic means for scope marking, and these are partly language specific. The capacity to relate utterances to each other in a coherent discourse encompasses the acquisition of language specific means in a whole range of domains. Learners pass from a lexical stage to a morphological stage of acquisition. For example, in order to overcome over-explicit (and sometimes even misleading) reference with full noun phrases in contexts of information maintenance, learners are pushed towards the acquisition of anaphoric pronouns as a cohesive device. In order to become independent of the principle of reporting events in their natural chronological order, learners are pushed towards the acquisition of morphological tense and aspect marking (Dietrich et al. 1995 and Starren 2001). The pronoun system and the morphological tense and aspect system are again very language specific. Earlier studies have shown that the acquisition of morphological markings obliges learners to embark on the insecure grammaticalization process (Klein & Perdue 1994). Some of the papers in this volume study learners that have already mastered the language specific form side of the target system. These near-native speakers are able to construct complex bodies of coherent discourse and to deal with a lot of language dependent properties at a time with help of morphological means. However, it is not only the form of the relevant language specific devices they have to master. Target-like forms that the learner acquires have to be used in target-like ways (with target-like functions). Languages exhibit considerable diversity in their morphological markings in that meanings which are coded in some languages in lexical form are mapped into grammatical form in others. Previous cross-linguistic studies have shown that grammaticised meanings play a crucial role in determining the decisions which speakers make in the selection and linearization of information (Carroll et al. 2000).





Introduction

Under this perspective, information structure also encompasses factors which determine the kind of information which is selected for expression and how it is mapped into linguistic forms. Native speakers of different languages have clear preferences for the encoding of concepts that are more easily expressible with the means provided by their languages (Slobin 1996). For example, speakers of languages in which a concept of progressivity is grammaticised on the verb (English, Spanish, Arabic for example) are more likely to express aspects of a dynamic situation as compared to speakers who have to resort to lexical means in order to convey the same meaning, such as in French or German. The papers by Carroll and Lambert, Murcia-Serra, and Watorek show that second language learners do not start from the base line, but that they draw upon the information organization principles favored by the grammatical preferences of their L1. So, at some point in L2 acquisition, routines of information organization seem to become a hindrance rather than a pushing factor in acquisition. The papers in this volume examine language-internal explanatory factors and in particular the communicative and structural forces that push and shape the acquisition process, and its outcome. On the theoretical level, the approach adopted appeals therefore both to formal and communicative constraints on a learner’s language in use, providing a ‘testing ground’ for the respective weight of grammatical versus functional determinants in the acquisition process. The following two empirical domains have proved to be highly relevant for testing the grammatical and functional determinants and are therefore addressed in the papers: (1) the expression of finiteness and scope relations at the utterance level and (2) the expression of anaphoric relations at the discourse level. (1) Finiteness or morpho-syntactic marking of the verb is strikingly absent from the basic stages of language acquisition and it is interesting to compare first and second language acquisition with the following questions in mind: what are the structural precursors of verbal morphology and how do they evolve in later stages of acquisition? Seen from a functional perspective, morpho-syntactic tense and aspect marking is generally attested as one of the most typical functions of verbal morphology. In order to investigate how children in L1 and adults in L2 acquire the complex devices necessary to anchor utterances in time, papers concerned with this question provide an in-depth analysis of the temporal features of learner utterances in their discourse context. A particularly interesting point is the interaction of the acquisition of verbal morphology with the evolving possibilities of integrating scope bearing elements (e.g. scope particles and negation) into the utterance. (2) The papers in the second part are concerned with the acquisition and use of anaphoric devices for reference maintenance in narrative and descriptive discourse types. This is particularly interesting for second language acquisition, since

Introduction

adult L2 learners can be assumed to know in principle how to construct coherent and cohesive discourse; still they are shown to have considerable problems in making the right choice with respect to the means for anaphoric linkage. The question then is how the development of further linguistic means enhances discourse coherence. Or conversely, whether the communicative need to enhance discourse cohesion pushes acquisition. As is shown in Gullberg’s paper, implicit discourse principles and non-linguistic devices like the use of gestures can, when combined, significantly reduce the need for explicit marking of discourse cohesion. A further, and important, question here is whether the way in which even very advanced L2 learners try to reconcile utterance-structure and the informational organization of text-types shows some influence of these learners’ first language. Overall, the book should make an important contribution to language acquisition studies for the following reasons. Firstly, the same questions are asked of child L1, child L2, child bilingual, and adult language acquisition. Secondly, these questions go far beyond many other attempts to compare L1 and L2 acquisition, which are restricted to the level of (morpho)syntax. Thirdly, the range of proficiency of adult language learners goes from the very beginnings of the process to quasi-bilingual, and, fourthly, the approach is resolutely cross-linguistic. The volume is thus dealing with a great diversity of testing grounds for the claim that the dynamics of language acquisition are located at the interface of information structure and the linguistic means the learner brings along.

Content of the volume The volume is divided into two major parts that correspond to the two empirical domains addressed in the previous section: “Finiteness and Scope Relations” and “Anaphoric Relations”. Papers in part one deal with the question how learners manage to acquire form and function of finiteness and to integrate scope bearing elements like negation, temporal adverbials and scope particles into the structure of their utterances. The papers in part two focus on the acquisition of languagespecific preferences for anaphora, cohesive patterns and means of information organization in narrative and descriptive discourse.

.

Finiteness and scope relations

The first part, which deals with the organization of information at the utterance level, contains four articles about the semantic, morphologic and syntactic characteristics of the acquisition of finiteness by child and adult language learners





Introduction

(Schlyter, Haberzettl, Dimroth et al., Gretsch), and four articles that investigate the integration of scope sensitive elements in relation to the acquisition of finiteness (Giuliano, Bernini, Benazzo, Hulk). The final paper in this section (Becker & Veenstra) shows how features that are important for the expression of information structure in a synchronic perspective of acquisition can also play a role in the interpretation of data analyzed from a diachronic perspective. Suzanne Schlyter investigates the development of finiteness in terms of verb morphology and syntax in adult Swedish speaking L2 learners of French in comparison to bilingual children acquiring French and Swedish as first languages from birth. With respect to morphology the author discusses different criteria for calling some form ‘acquired’. In order to see whether the learners have access to finiteness in a syntactic sense (IP), negation, subject- and object-pronouns, case and subordination are considered. The most important difference between child and adult language acquisition consists of the fact that children, after a complete lack of verbal morphology or other evidence for the instantiation of IP, acquire morphology and syntax simultaneously, and make use of morphological cues to build up their syntactic structure, whereas the adult learners already show signs of access to functional categories at the earliest stages of acquisition. Their verb morphology however is deficient and variable until much later in the development. Other than children, adults can rely on their already acquired syntactic knowledge but have to learn how to use the specific word forms of L2. Stefanie Haberzettl studies the first steps in the acquisition of verbal morphosyntax by investigating a structural pattern attested in the speech production of two Turkish speaking girls learning German as a second language. This pattern combines the copula or auxiliary ist and a lexical verb, mostly an infinitive. The author investigates several functional explanations for the non target like ‘ist support’ and concludes that there is no such function to this idiosyncratic pattern: the patterns with discontinuous verb forms (ist with lexical infinitive or past participle) do not cover other functions than utterances without ist, e.g. with respect to aspect marking, but represent pure formal variation. The author therefore proposes a form-oriented explanation of the data, where the idiosyncratic ist pattern is interpreted as a creative construction based on patterns detected in the input. This early form-oriented ‘tinkering with chunks’ proves to be useful at later stages of acquisition since it serves as a formal precursor for target-like verb placement in periphrastic constructions (discontinuous position) and embedded sentences (verb final position). Christine Dimroth, Petra Gretsch, Peter Jordens, Clive Perdue and Marianne Starren study the acquisition of finiteness in children acquiring Dutch or German as their first languages and adults acquiring these languages as their second languages. The authors distinguish between the semantic concept of finiteness and

Introduction

the way languages (often morphologically) mark it. They argue that finiteness is the carrier of assertion in an utterance and relates the descriptive content of a sentence to its topic component. It is shown that child and adult learners rely on similar means for the expression of this concept at subsequent stages of acquisition. At the so-called ‘Conceptual Ordering Stage’ i.e. before target-adequate morphological markings become productive, learners establish the assertive relation by a closed class of linking elements which contains elements expressing positive or negative assertion, scope particles, and modal phrases. At the ‘Finite Linking Stage’ assertion marking grammaticalizes. Elements of the target functional category of auxiliaries come to be used as a grammatical linking device whereas scope particles and other target adverbial elements do no longer occur as independent linking elements. Whereas the illocutionary linking elements of the ‘Conceptual Ordering Stage’ are adjuncts, auxiliary verbs are part of a functional category system. As is the case in the target language, they function as the head of a head-complement structure at the ‘Finite Linking Stage’. The paper by Petra Gretsch is concerned with the development of temporality in L1 as compared to L2 acquisition. It is generally assumed that L1 learners follow a morphological strategy, which means that early morphology is supplemented with late adverbials. L2 learners, on the other hand, are to follow an adverbial strategy, where early adverbials get complemented with morphological markers at later stages of acquisition. An analysis of temporal anchoring in self-initiated narratives from three longitudinal German child corpora reveals that the morphological strategy and the adverbial strategy can emerge simultaneously in L1. Which kind of strategy is used first depends on the paths individual learners follow, but there is no radical different path of access to grammar between L1 and L2. For those children who prefer morphology over adverbials the question arises why they chose to go the more complex way. The author suggests that the morphological strategy has in fact one advantage over the adverbial one – at least for children. Children’s basic conceptualization of the temporal space and its gross partitioning in here&now vs. distant from here&now is directly reflected in the two-way split of early verbal morphology (finite vs. infinitival). Since temporal adverbials do not occur as salient opposition pairs in the same way as the verbal morphology does, they are less qualified to represent the child’s early partitioning of the temporal space. It is thus rather the meaning than the structural integration of adverbials that might lead some children to chose the formally complex morphological strategy. The paper by Patrizia Giuliano is concerned with the acquisition of negation, verbal morphology and the relation between these two domains in L2 French and English. The correlation of relatively early finiteness marking on non-thematic predicates and the emergence of post-verbal or post-auxiliary negation with exactly these verbs is given an explanation which hinges on the relational nature of these





Introduction

verbs. Relational predicates play a complex role in the development of learner varieties since they are typically the first grammatical means used to mark tense and aspect. The author attributes their precocious acquisition to their relatively accessible morphology and to the fact that auxiliated constructions allow for a useful division of labor: the auxiliary marks features of finiteness (i.e. tense and/or aspect and/or agreement) whereas the thematic verb carries the lexical information. This also explains the observation that the acquisition of post-verbal negation is sensitive to productive morphological oppositions on relational predicates. Negation cannot have scope over the tense information encoded in relational predicates, because the non-validity of the utterances lexical content needs to be asserted for a given time span. In the basic stages of L2 acquisition, scope differences of this sort are expressed in a transparent way through different surface positions of negation. As a consequence, the negator follows relational predicates from early stages onwards, but precedes thematic verbs on which temporal markings are lacking or ambiguous. Giuliano Bernini addresses similar questions with respect to the copula in L2 Italian. The copula is systematically missing at the basic stages of acquisition despite the fact that it is obligatory in the learners first language, Tigrinya. On the other hand, it is the first verbal element to develop finiteness, tense distinctions, and the target-like post-finite position of negation at later stages of acquisition. Bernini shows that forms of essere function as independent finiteness operators that can be combined with nouns, adjectives, participles or other verbal elements which, in the learner variety, cannot yet incorporate the expression of finiteness. With the early possibility to code tense distinctions, the copula can also be used as means of enriching the morphological potential of post-basic stages. When combined with other verb forms, its imperfective forms substitute for imperfective morphology of verbs with non-durative Aktionsart or other verbtypes in the period where imperfect morphology of full verbs has yet to be developed. The early post-finite placement of negation attested with the copula is ascribed to the fact that, at that stage, scope marking is still tied to surface order and both, finiteness and tense, which are encoded in the copula, are outside the scope of negation. Sandra Benazzo investigates the interaction between the development of verb morphology and the acquisition of items such as again, already, still, yet, which are referred to as ‘temporal adverbs of contrast’ (TACs). Though temporal adverbs have been shown to play an important role for the early expression of temporality, TACs are acquired relatively late. Based on an analysis of longitudinal data of English, French and German L2, the author shows that only items expressing the reiteration of an event (such as again) are productively used at the Basic Variety stage, while other TACs (such as already, yet, still) are only attested at more advanced stages of acquisition, usually when the verb is functionally inflected.

Introduction

Instead of explaining this late acquisition on the basis of the cognitively complex temporal configurations expressed by TACs, the author suggests, that it is the development of verb morphology itself that determines the possibility to integrate TACs. The association of these items with a functional verb morphology is necessary to get their standard temporal meaning: already marks the transition from a negative to a positive phase of a state or an event, while still marks its continuation from a previous undetermined time interval. In both cases finite verb morphology indicates which is the actual time span of the event talked about, while TACs make reference to alternative (previous) time spans of the same event. Aafke Hulk studies scope-taking discourse particles in bilingual L1 acquisition of Dutch and French as compared to the monolingual acquisition of these languages. The French particle aussi (also) and its Dutch equivalent ook are considered to be bridging elements between the containing utterance and the preceding discourse. Together with syntax, pragmatic information is needed to determine which element of the containing utterance is actually affected by the particles. The acquisition task therefore comprises both, the linguistic devices and the relevant principles of information organization in discourse. In bilingual acquisition, elements involving this kind of interface between syntax and pragmatics have been shown earlier to be particularly prone to intraindividual cross-linguistic influence: a construction shared by both languages that is possible but less frequent in the monolingual child version of one of the languages is then overused by the bilingual child. This was confirmed for the acquisition of aussi and ook. Besides different positions in monolingual child Dutch and French there is an utterance internal position that is attested early in both languages. Still it is less frequent in French than in Dutch, where ook in that internal position has been shown to function as a precursor of functional categories of the verbal domain. The pioneering function for syntactic structure building that ook has in monolingual child Dutch might have encouraged the bilingual child to use the position more often with French aussi than monolingual French children do. The article by Angelika Becker and Tonjes Veenstra is concerned with the question to what extend properties of verbal morphology in French related Creoles are attributable to processes of second language acquisition of French involved in their creation. Processes of untutored adult L2 acquisition took place when the first generation of imported slaves came into contact with French as a superstrate language, and the authors assume that the outcome was similar to the Basic Variety (BV) attested for modern learners of French. With respect to verbal morphology, French BVs show free formal variation of long and short forms without functional distinctions. While that long/short variation gets a target-like functional correlate in more advanced stages of second language acquisition, French related Creoles seem to have assigned different semantic correlates to the free formal vari-





Introduction

ation that was characteristic for the learner variety spoken by the first generation of Creole creators. Most interesting are cases like Morisyen, where a target shift occurred due to early superstrate-withdrawal. In a process of bilingual L1 acquisition the crucial second generation developed two native languages, their parents native language and the BV created by the first contact generation. When the target shift occurred, the formal variation in the BV did not yet have a functional correlate. The second generation fitted the formal variation into a new paradigm and passed that language on as the (only) L1 of the third generation.

. Anaphoric relations The papers in part two focus on the acquisition of language-specific preferences for anaphora, cohesive patterns and means of information organization in the domains of time, entities (Carroll & Lambert, Murcia-Serra, Gullberg) and space (Watorek) in narrative and descriptive discourse. The paper by Mary Carroll and Monique Lambert is concerned with the question, why native speakers narratives sound native-like and those of second language learners, though formally correct, do not. The authors study film retellings produced by advanced French and German learners of English and reveal that the way in which these learners construct their narrative discourse still shows traces of the grammatical preferences for information selection and organization that are operative in their L1. The German learners of English are shown to have difficulties with a reorganization of the core factors which guide the type of information which is selected for mention. In deciding what to say at the level of macro-planning, they do not direct attention, for example, to inanimate entities to the degree found in English and French, where the syntactic subject plays a prominent role (through either fixed word order or pro-drop properties) and attention in information selection is directed to all types of entities. Speakers of French and German (languages, in which progressive aspect is not grammaticised) have problems in recognizing the functional role of a distinction such as the simple tense versus the progressive in narratives in English, its overall status in information structure and with this its role in a clearly defined ‘system of meanings’ or form-function relations. The paper by Jorge Murcia-Serra presents a study on the acquisition of the linkage between syntactic relations (especially subject), semantic roles (primary agent) and the informational status of referents in narrative texts. This linkage, which varies in Spanish (the learners’ L1) and German (the learners’ L2), is considered to correspond to different patterns of conceptualization, namely attention

Introduction

focused on agent entities vs. attention focused on a primary topic entity. The analysis problem for Spanish learners of German is not only the identification of the relevant linguistic structures for marking reference in narratives in the target language but also the realization of the corresponding conceptual patterns. The results obtained show that the use of the category subject as a coding for the global topic entity required in the target language seems to pose important problems for learners of German with L1 Spanish. The acquisition of such a linking not only requires the recognition of the function of the category subject for textual cohesion but also, in the case of narratives, the focusing of attention on the protagonist of the events being presented. At the same time, it is necessary to diminish the attention given to actor entities in the source language. This means, that the problem is not the acquisition of the formal means, but basically one of the appropriate usage of such means paired with the corresponding patterns of conceptualization required for speaking in the target language. Marianne Gullberg addresses the question how L2 learners’ co-speech gestures can shed light on certain characteristics of L2 speech production, especially the well-known tendency of overexplicit anaphoric reference to maintained information that is often attributed to the avoidance of complex pronominal systems. Gestural anaphoric linkage normally involves the anchoring of referents in some part of space at first mention. The L2 learners studied by Gullberg over-mark referents gesturally in that they anaphorically indicate them at every mention. But they only over-mark in gesture what is also over-marked in speech by full nominal expressions. Two explanations for this observation are discussed. Firstly, gestural anaphoric linkage could well have a communicative motivation in that it allows unequivocal visual co-reference. This might be especially helpful for L2 learners, since their over-explicit speech can also lead to ambiguity, and is sometimes difficult to process for the addressee. Another explanation would be that these gestures reflect ‘idea units’ underlying learners’ speech planning. Overuse of anaphoric gestures could then be interpreted as reflecting learners’ L2 speech planning which proceeds by smaller units, since even reference to maintained entities needs to be planned separately. Marzena Watorek investigates anaphoric relations in the domains of space and entities in descriptive discourse. The author presents the results of a cross-sectional and cross-linguistic study involving intermediate and advanced Polish learners of French as well as native speakers of both languages. Polish and French differ with respect to the preferred patterns of spatial organization as well as with respect to the available means for encoding spatial relations. Differences in structure and semantics of the preferred patterns also have an impact on the strength of discourse cohesion (anaphoric linkage).





Introduction

A comparison of both groups of learners reveals a complexification of the spatial concepts expressed. In the less advanced learners, an over-use of a minimal phrasal pattern based on the existential verb il y a (there is) and an omission of explicit reference to space is attested: speakers localize as many entities as possible in relation to a salient anchorpoint which they can implicitly refer to in many subsequent utterances. If need arises, discourse cohesion is reinforced with help of additive scope particles which indicate explicitly that reference to that anchorpoint is maintained. With increasing proficiency, reference to these spatial anchorpoints becomes more explicit and more locative verbs are observed. These locative expressions then also play an important role for anaphoric linkage, and pose some problems for the Polish learners of French, who show an over-use of demonstrative forms in contexts of immediate maintenance of the NP-relatum.

References Carroll, M., Murcia-Serra, J., & Watorek, M., & Bendiscioli, A. (2000). The relevance of information organization to second language acquisition studies: the descriptive discourse of advanced adult learners of German. Studies in Second Language Acquisition 22(3), 441–466. Dietrich, R., Klein, W., & Noyau, C. (1995). The acquisition of temporality in a second language. Amsterdam: Benjamins. Klein, W. & Perdue, C. (Eds.). (1992). Utterance Structure: Developing Grammars Again. Amsterdam: Benjamins. Lambrecht, K. (1994). Information Structure and Sentence Form. Cambridge University Press. Rooth, M. (1985). Association with Focus. Ph.d. dissertation, University of Massachusetts. Amherst. Slobin, D. (1996). From thought and language to thinking for speaking. In J. Gumperz & S. Levinson (Eds.), Rethinking Linguistic Relativity (pp. 70–96). Cambridge: Cambridge University Press. Starren, M. (2001). The Second Time: The Acquisition of Temporality in Dutch and French as a Second Language. Utrecht: LOT. Vallduví, E. (1992). The Information Component. New York: Garland.

JB[v.20020404] Prn:28/04/2003; 10:17

F: SIB26P1.tex / p.1 (36-61)

P I

Finiteness and scope relations

Development of verb morphology and finiteness in children and adults acquiring French* Suzanne Schlyter

.

Introduction

In this study, the development of finiteness in French in terms of verb morphology and syntax will be studied in adult Swedish-speaking L2 learners of French, in comparison to bilingual Swedish-French children acquiring French as one of their languages from birth. In studies of second language development, an important question is whether there is a relation between the development of verb morphology and that of syntactic phenomena such as negation, word order, etc. Scholars have obtained different results, which partly depends on different definitions of “the acquisition of finiteness”. In this study, such accounts will first be presented, from more morphologic to more syntactic definitions, in order to show the complexity of the question, especially as concerns adult L2 learners. It will then be argued that in the child learners, after an initial stage of no syntactic or morphologic evidence of finiteness, such evidence appears at a certain age (around 2;2–2;6) at which morphologic and syntactic criteria coincide. In contrast, in adult L2 learners of French, finiteness in the sense of syntax must be dissociated from finiteness in the sense of morphology: the L2 learners studied can be said to have access to finiteness in a syntactic (and certainly semantic) sense from very early on, but develop a systematic and targetlike morphology of the different verbs only very late, after many years of residence in France. In order to see whether the learners have access to finiteness in a syntactic sense, negations, subjects, object pronouns and subordinations will also be considered, but more superficially. Differently from Dimroth et al. (this volume), who use a semantic and functional framework, this study will build on formal, morpho-syntactic categories.



Suzanne Schlyter

. Finiteness, functional categories and morphology Finiteness can be defined semantically, as expressing assertion (Klein 1998), syntactically, as an abstract Functional Category responsible for certain syntactic phenomena, or morphologically, as being expressed by finite verbs. In generative syntax, finiteness is related to the functional category INFL (= Inflection).1 In this paper, I will not go into the details of generative syntax, but will use some of its more widely accepted results in order to discuss child and adult acquisition data. As for the acquisition of finiteness in a first language, L1, it has long been shown that there is a relation between the development of verb morphology and certain syntactic phenomena. Clahsen (1982) showed, for German, that when children were able to distinguish different forms of subject-verb agreement such as ich gehe / du gehst / er geht in their production, they also correctly used the V2 word order, inverting subject and finite verb after a preposed element. Similar findings have been obtained for French (Meisel 1994; cf. also Pierce 1992), showing that when the children had arrived at a stage where they distinguished between 1st and 2nd or 3rd person singular (with subject clitics + finite verb), they were also able to place the negation (pas) correctly after the finite verb and before the non-finite one, indicating a relation between agreement and verb placement. Regarding syntactic structure, these analyses show that when INFL is accessible for the child (see Figure 1 below), the main verb can move to it, which results in a postverbal negation, also with lexical verbs. CP COMP

IP’ SpecIP

IP INFL

Neg P Neg

VP V

que that

X X

a has

X X

lit reads

pas not

NP

lu read

le livre the book

– –

le livre the book

Figure 1. Syntactic structure (simplified) of French declarative negative sentences, and raising of the main verb

Development of verb morphology and finiteness

Most authors agree that, during L1 development, L1 learners proceed in some sense “bottom up”, i.e. they first use lexical elements of VP (read book), later adding morphological elements such as auxiliaries, i.e. instantiating IP+VP (X has read the book) and finally using subordinations, i.e. CP+IP+VP (that X has read the book), but do not produce, e.g., CP+VP without IP (e.g., *that read book). Certain authors (Clahsen et al. 1993/1994; Meisel 1994), account for such facts using the hypothesis of Structure Building from the bottom up, i.e. in the order VP – IP – CP, whereas others account for similar facts by proposing truncation of the syntactic tree below CP (leaving IP and VP intact) or below IP, leaving VP intact (Rizzi 1993/1994). In second language acquisition, a relation between verb morphology and syntactic facts is less evident and the positions differ. Klein (1986) proposed, in his state-of-the-art work on studies of negation in L2 acquisition of different (essentially Germanic) languages, a relation between what he called “finiteness” and the place of the negation. It was proposed that when learners were acquiring a more developed morphology, they abandoned an earlier pragmatic positioning of the negation and replaced it with a syntactically defined place, directly after the finite verb. Clahsen and Muysken (1986) and Meisel (1991) argue for a clear difference between L1 and L2 acquisition as regards the relation between morphology and syntax. Meisel resumed these studies showing that whereas children rapidly acquire agreement, without errors, L2 learners do not actually acquire agreement but instead make many errors, and the position of the verb is not dependent on the +/–finite distinction. The relation between finiteness and the place of negation was further questioned by Meisel (1997), who argued against such a relation in L2 acquisition, whereas Parodi (1998, 2000) and Giuliano (2000), following her, show that such a relation exists if we restrict the development of verb morphology to non-thematic verbs (i.e. be, have and modals). Parodi (2000) argues that the nonthematic verbs do not have semantic content, but are spell-out (or placeholders) for syntactic features like agreement, tense and modality, and are directly generated in IP, whereas thematic verbs have semantic content, are generated in VP, and must move to INFL in order to be placed before the negation. In recent years, the lack of relation between morphology and syntax in L2 acquisition has been further discussed. The “Missing Surface Inflection Hypothesis” (Prévost & White 2000; cf. also Lardière 1998) states that adult L2 learners may exhibit syntactic signs of having access to IP/INFL, but still do not use verb morphology correctly. Lardière shows that the learner she studies, in spite of not having developed tense marking, does use, e.g., case marked pronouns (which can obtain their case only in IP) and other syntactic phenomena derived from the presence of INFL. Prévost and White show that negation in French L2 is correctly placed after the finite verb from rather early on, that subordinate clauses – also without finite verb forms – are used early, and they argue that the problems with verb morphology result from learners using a default (nonfinite) form in finite positions, but





Suzanne Schlyter

not the opposite. This is taken as an argument for adult learners’ early access to IP (and CP), whereas the problems of verb morphology are more superficial phenomena, showing variation over an extended period and possibly never being acquired. Such a position is related to the “Full Access – Full Transfer” position (Schwarz and Sprouse 1996), according to which syntactic structure is considered as being taken over from the source language, and not gradually built up, as is proposed by many other theoretical positions. Scholars who have compared L1 and L2 acquisition agree that bound morphology is acquired early and without many errors in L1, whereas adult L2 learners have great difficulties with this, and often replace bound morphemes through compensatory lexical strategies. Free grammatical morphemes are normally reported as being somewhat easier to acquire (Wode 1981; Parodi 1998).

. Varying acquisition criteria If we wish to relate syntax to the acquisition of verb morphology or to compare the appearance of finiteness in L1 and L2, we have, in both cases, the problem of defining an exact point at which morphology is acquired, or at which learners have access to IP; establishing these points can be very problematic (cf. Jansen 2000). In the 80s, when morphology, finiteness and the category AUX (or INFL) were often seen as more or less identical, many scholars used morphological opposition between forms of the same verbs (e.g., je parle / pour parler), or a productive use of morphemes in nonsense verbs or in verbs invented by the learners, as criteria for the acquisition of this entire complex. Correct verb forms, but with different verb types, such as ich gehe vs. hab gemacht, or je parle vs. pour manger, were not necessarily taken as arguments for productive morphology, since they could be seen as non-analyzed chunks (cf. Pienemann 1998). According to such a view, suppletive forms of a paradigm such as je suis vs. il est, je vais vs. il va could be excluded, if considered as rote-learned chunks and not as productive morphology. These then constituted very restrictive criteria for acquisition of morphology. Thus, if finiteness was related to acquisition of verb morphology, then the criteria for finiteness were necessarily strongly restricted, and finiteness necessarily appeared late.2 However, if finiteness was not only purely morphologically defined, but considered as identical to the category INFL (= Inflection) or one of its subcategories, then syntactic criteria would also be taken into account. Thus, later studies indicating the correct placement of negation after copula and auxiliary verbs, even if the lexical verbs had not yet developed correctly, postulated much earlier access to INFL. Parodi (1998, 2000) argues for access to IP by her adult L2 learners of German at an early stage, during which they produce non-thematic verbs in morphological opposition (ich bin, du bist etc.), even if they do not yet inflect the lexical,

Development of verb morphology and finiteness

thematic verbs. Hawkins (2001: 62) goes a step further, arguing that the mere presence of the copula in L2 English is evidence for IP, since it always occurs before the negation. If one has decided on the type of criteria to be considered as relevant, according to the discussion above, the next problem is the quantitative criterion: should we count from the first appearance of a form, from 30%, or use the old criterion of 90% correct use in appropriate contexts? Jansen (2000) discusses the need for stricter acquisition criteria, and both she and Hawkins (2001: 333) show how different conclusions are drawn from the same data using different criteria. In this way we can observe that, as concerns French, Klein and Perdue (1997) argue that the Basic Variety has “practically no inflectional morphemes” and that finiteness is developed only in the post-basic stages, whereas Prévost and White (2000), using partially the same data, argue that INFL and all functional categories are there from the start. From these discussions, it seems that any theoretical position can be supported using any kind of data. In this way, the following criteria (among others) can be used to establish the time in the development at which finiteness, or IP, is acquired: A. independent of morphology: the first syntactic traces of IP (postverbal negation, case, CP, etc.); B. the first appearance of copula, i.e. c’est; C. inflection of copula and auxiliaries; D. correct functional opposition of some few different forms of the lexical verbs; E. correct functional opposition of almost all different forms of the lexical verbs. It is evident that the postulated time point of acquisition will be very different, at least for L2 learners, depending on which criterion is used. In this study, these different criteria will be discussed, and it will be shown how they lead to different results for child L1 acquisition and for adult L2 acquisition. Since finiteness is a notion having a great variety of definitions, we will henceforth speak only about Verb morphology, on the one hand, and the syntactic category IP, or INFL, on the other, the latter being abstract categories postulated for certain syntactic phenomena. It is not impossible that INFL might have a semantic correspondence in the semantic category FIN (Klein 1998), but the definition as used here is purely syntactic.

. Verb morphology, and syntactic evidence for IP in French Before discussing the acquisition of verb morphology in French, it is necessary to consider the actual spoken system of the target language, which is very different from its written counterpart. Written French, with its conservative spelling, is a





Suzanne Schlyter

strongly suffixal language (je parle, tu parles, il parle, nous parlons, vous parlez, ils parlent; Passé Simple il parla, Futur Simple il parlera etc.). Spoken French has lost most of the suffixes, and can rather be considered as developing a prefixal morphology, with auxiliary instead of suffix for past and future tense (Passé Composé il a parlé, Futur Proche il va parler), and subject clitics as possible person prefixes (see below). The only verb suffix that is still productively and frequently used in informal spoken varieties of French (including early child language and learner language) is the Imparfait suffix: était, avait, voulait, habitait, aimait etc. But also this form has a tendency to be lexically restricted, and in spoken informal French it is most frequently used with copula, auxiliaries and modals. The forms Conditionnel and Futur Simple exist in spoken language, but are productively used only with some frequent verbs (voudrais, sera, aura, fera, etc.). Many scholars attribute to the subject clitics an affixal or quasi-affixal status (cf. Auger 1995) – which is compatible with the general pattern of prefixal morphology. This means that je and il in, e.g., “Moi je parle” and “Le prof il est sympa” are markers of subject-verb agreement, moi and le prof being considered (under certain circumstances) as the subject, and not as a dislocated element. In this way, the subject pronoun in informal spoken French may have the status of a person affix (je /parl/, tu /parl/, /iparl/, on /parl/, /iparl/, etc.), not very different from Italian and Spanish suffixed person inflection (parlo, parli, parla, etc.). I will, therefore, include them here among possible morphological elements. However, the system is variable and not quite established; traditional forms (e.g. nous parlons) occur in more formal registers, in which most of these pronouns may also have a nominal, and not suffixal, status. As for the L2 learners studied here, they do not seem to have captured the morphological status of the subject pronouns, but use them as nominal pronouns (Granfeldt and Schlyter 2001, 2003). The most productive opposition of suffixal bound morphology in familiar spoken French can instead be said to be that between the ‘short form’ (/parl/) used for present tense, i.e. the most frequent finite form, and the ‘long form’ (/parlE/) used for infinitive (= parler) and past participle (= parlé). The latter may be perceived by the learners as a general form for nonfinite verbs (cf. Klein, Dietrich, & Noyau 1995, on this essential opposition.) The long form can also represent the Imparfait (= parlait). However, normally the Swedish learners do not use Imparfait until very late, so generally in early stages, and clearly when they refer to present time, such forms can be considered as infinite forms. Very productive free morphemes are those that mark tense/aspect, namely ai/a(s) (have), suis/es(t) (be), vais/va(s) (go), used respectively for the most frequently used past tense, Passé Composé: il a parlé, elle est arrivé etc., and for Futur Proche: on va partir. These are, besides the subject pronouns, the most important and frequent verb morphemes in spoken French, and must be taken into account in a study of verb morphology. Modal verbs (veux, peut, faut etc.) have a function

Development of verb morphology and finiteness

similar to that of the auxiliaries just mentioned. Even if they are not traditionally considered as morphological elements, I will consider them as such here, just as I will the auxiliary verbs. The reason for this is that all these elements, in their most unmarked form (a, e, va, veu, peu, etc.), are used in very early stages, both by children and by adult learners, as markings of TMA (tense, modality, aspect). These auxiliary elements also have their own inflection, through suppletive forms: j’ai/ il a/ ils ont; je suis/il est/ ils sont; je vais/ il va/ ils vont etc. It is discussed in the literature whether to consider them as morphological variants or as separate lexical entries (see above). French auxiliaries are apparently somewhat different in this respect from German or Italian ones, which have productive morphological inflection to a greater extent. However, as Prévost and White (2000: 120) point out, if learners use such forms in French correctly, it implies that they have access to INFL,3 since they must check for (use the correct form for) different grammatical persons. These forms will be taken into account here, but their interpretation is not straightforward. As for syntactic indications of the presence of INFL in French, we cannot use word order such as the German V2, but we can use the postverbal placement of the negation pas, and also of sentence adverbs like toujours, souvent etc., since French is a language in which the main verb raises to INFL, as shown above. We can also use the case criterion, as mentioned above, since French pronouns – if we consider them as nominal elements – distinguish nominative case (je/tu/il/ils) from accusative/dative case (me/te/le/la/lui/les/leur) and from oblique case, used after prepositions (moi/toi/lui/eux – the same forms as the strong pronouns). The object clitics (je le vois), which have to be moved to their position from (or possibly agree with) a nominal postverbal position (je vois le chat), are another indication of the existence of a Functional Category above VP, see Granfeldt and Schlyter (2003).4 More indirect evidence for the existence of Functional Categories such as INFL is the presence of complementizers, COMP (que, qui, quand, où, si etc.). According to both the “truncation theory” and to “structure building theories”, learners cannot have access to COMP if they do not also have access to INFL (cf. above). We have, therefore, in this study also taken into account the presence of COMP in the data from both groups. The occurrences of these types of morphology as well as these syntactic criteria according to their first appearance will be compared in the L1 learners and the adult L2 learners in Sections 3 and 4 (Tables 1 and 2) below. The categories studied are: postverbal negation (fV+neg), case (je vs. moi etc.), preverbal object clitics (ocl), subordinations with subjunctions (C), nonfinite verb forms (Nonf), subject pronouns and verbs (“scl+V” in children, “Psuj+V” in adults), auxiliaries (Aux+V), modal verbs (Mod+V), and Imparfait (Impf). Morphological variation and its gradual reduction, leading to finiteness according to criterion D above, will be accounted for in Section 5 (Tables 3 and 4) below.





Suzanne Schlyter

. Learners studied This study is part of a research project at the University of Lund on the development of functional categories in L1 and L2 French. In order to have comparable data, where the possible influence from another language is controlled for, we have chosen to compare bilingual children, growing up with both Swedish and French simultaneously (henceforth (2)L1 learners), and adult Swedish learners acquiring French. The following learners are studied in the project:

Bilingual children: The bilingual Swedish-French children (Anne, Jean, Mimi, Dany) were recorded about once a month between 2 and 4 years; the children were middle class, they all lived in Sweden, growing up with the ‘one-person-one-language strategy’. They were recorded at home with each of their parents and in the presence of a Swedishspeaking assistant. Two additional children were studied in the project, but since their French was very weak, they will not be considered here. Adult Swedish-speaking learners of French: The adult learners were around 20–40 years old and were students or artists. Most of them had finished secondary/high school, knew English and in a few cases another language. Some of them acquired French in a natural setting, during some years of study or work in Paris (Martin, Petra, Sara, Karl, Johan and Knut), but did not (or minimally) attend courses in French. These are referred to as the Informal Learners. Others had acquired French in school (for 6 years, about 500 hours including homework) and were during the recording period studying French full time at university, in the first semester (Lisa, Sama, Nina and Ylva). They had not been in a French-speaking country for more than one or two weeks. These are referred to as the Formal Learners. The least advanced learners (Sara, Petra) had been in France for about 3 to 5 months at the onset of the studies. In order to obtain data from an even more elementary level, one recording was made with Henry, an adult who had never previously learned French. He had heard some French during holidays in France, could understand some written French in newspapers etc., but did not speak French since he had never had to communicate in that language (an input roughly corresponding to about 3 months). In this study, we are most interested in the L2 learners and their development of morphology and finiteness, but data will also be drawn from the bilingual L1 learners, for comparison.

Development of verb morphology and finiteness

. First language acquisition: The bilingual children In our data, in the very first recordings, there are no surface realizations of verb morphology in any of the senses discussed above (see criteria B–E above), nor any syntactic traces of IP, which makes it plausible to propose that the children do not yet project or lexically instantiate INFL. Quantitative data from the four children are presented in Table 1 (page 27) below, and to illustrate, examples from one or two children are presented. Our first example is from the child Anne, whose French is somewhat stronger than her Swedish.

. The first stage – no verb morphemes, no signs of instantiations of INFL In the first recording (A1, age 2;3), there are neither subject clitics, nor auxiliaries, nor opposing verb forms (in the sense that the same verb type occurs in different forms). As for syntactic criteria for finiteness, there are no postposed negations. Examples of her utterances: (1) a.

pleure, le bébé weeps, the baby b. donne! give! c. fait dodo make sleep ‘is sleeping’ d. fini! finished

(Anne 1, 2;3)

The verbs are used in appropriate contexts – present forms referring to ongoing events, participles such as fini directly after an event, and imperative (donne) before the desired event. This is often the case in child language, but not necessarily: if the child is creative and speaks a great deal, forms and functions may not be completely related at this early stage (cf. the child Ivar vs. the other children in Schlyter [1990], where this relation is studied systematically). Anne uses no negation in this first recording. A similar pattern can be observed in the earliest recordings of Jean and of Dany, see Table 1 (but not in Mimi, who is too advanced at the start of the recordings). Jean produces some negations, but these are not postverbal: (2) pas là not there

(Jean 2, 2;0)

Jean and Dany also use many non-finite verbs (Root Infinitives, cf. Rizzi 1993/1994; also Granfeldt 2003): parler, courir, parti etc. These are used without subject pro-





Suzanne Schlyter

noun or auxiliary, but according to the interpretation (i.e., recasts by the interlocutor), such elements can be considered as omitted: (3) Mother: quand est-ce que tu mets ton chapeau, Jean? ‘when do you put your hat on, Jean?’ Jean: courir run-inf Mother: courir. quand on va courir. run-inf. when you will run-inf. fini! finish-past ptc Mother: qu’est-ce que c’est que tu as fini? what is it that you have finish-past ptc ‘what have you finished?’ Jean: du lait ‘milk’

(Jean 2, 2;0)

(4) Jean:

(Jean 2, 2;0)

Children also often use nonfinite forms as imitations of an adult’s utterance, such that the subject and the modal verb are omitted in the child’s speech. Such imitations are not considered here. The way the verbs are used here, mostly uninflected and without subject (cf. Root Infinitives), is clearly different from the way the adult learners use the nonfinite verb forms, which are in most cases preceded by a subject in some form (see examples (16), (21), (27a, b), (28), (30a) below). This difference has also been pointed out by Prévost and White (1999).

. The second stage – verb morphology and syntax appear: INFL is instantiated In the second recording (A2, 2;6), some elements of verb morphology appear, which can be taken as instantiations of INFL: Anne uses different subject clitics here, which, as we mentioned above, can be seen as affixes of grammatical person; she also uses est as copula and as auxiliary: (5) a.

il est là! ‘he is there’ b. elle est cassée? ‘she is broken?’ c. je veux (celui-là) ‘I want that one’

Development of verb morphology and finiteness

d. je mets là? ‘I put there?’ e. c’est fini it-cl-is finish-past ptc ‘it is finished’

(Anne 2, 2;6)

According to Meisel (1994), the use of different subject pronouns indicates that the child manages the opposition of person morphemes, and in his data this correlates to the postverbal position of Neg. This indicates access to the INFL to which the verb is moved. Anne now uses postverbal negation, not only after est (a–b), but also after lexical (+thematic) verbs (c). (6) a.

c’est pas xx ‘it is not xx’ b. n’y est pas neg-there is neg ‘(it) is not there’ c. n’aime pas celle-là neg-like neg that one ‘(I) do not like that one’

(Anne 2, 2;6)

Such a stage is found also in the other children: Jean 4 and 5, and Mimi 2, are similar, see Table 1, below. Both non-thematic and thematic verbs appear early with postverbal negation. Later still, the verb morphology is evident, and different forms of the same verb are used, in correct opposition to each other (cf. criterion D above), see (7)–(8): (7) Situation: a puzzle falls to the ground Anne: oh, tombé! c’est tombé. oh, fall-past ptc it is fall-past ptc ‘oh, (it) fell! it fell’

(Anne 3, 2;8)

(8) Other situation: mother and child are reading a book about a little girl who is sitting on a chair and starts eating Anne: elle tombe! she fall-prs ‘she is falling’ Mother: non elle ne tombe pas. (. . . ) elle renverse la tasse. no she neg fall neg she knock-prs-over the cup’ ‘no she is not falling. she is just knocking the cup over’ (Anne 3, 2;8) Modals and auxiliaries start to appear:





Suzanne Schlyter

(9) veux tourner ça will turn that

(Anne 3, 2;8)

The negation (now in the more productive, single form) is placed after finite verbs, also lexical verbs: (10) a.

aime pas like-prs neg b. estime pas appreciate-prs neg c. aime plus like-prs no more

(Anne 3, 2;8)

Soon after, there are indications of case oppositions, such as je vs. moi and il vs. le: (11) a.

j’aime bien grenouilles I-nom-like well frogs b. non c’est moi no it-is me-oblique c. je le mets dans l’eau I him-acc put in the water d. il va nager? he-nom will swim?

(Anne 4, 2;10)

The children have certain difficulties with agreement between the subject clitic and the form of the verb (see also Ferdinand 1996): no child uses je suis during the time covered in Table 1, but only at about 3;5–4;0 years. Before this, though mastering je perfectly with other verbs (cf. (5c, d) etc.), they avoid je suis or use instead i e or something similar. Also j’ai seems to cause some problems: it appears late and is in the first time replaced by j’a, i a, je a etc. Except for Mimi (who is precocious in many respects and has all of these phenomena already in Mimi 2), the children do not oppose il a / j’ai until the age of 3;1 or later. This is clearly different from the adult learners, who make this opposition very early, cf. ex. (25) below. The children are very correct in their use of verb morphology. From this stage on, it seems that they never refer to past events with present forms, or inversely, and after a subject clitic the verb always has the finite form (je parle), never incorrectly the nonfinite one (*je parler). This is a confirmation of similar findings from many previous studies on French child language.

Development of verb morphology and finiteness

Table 1. Evidence for IP and CP in bilingual children, from syntax and verb morphology fV+Ng

Case

Ocl

C

Nonf

Scl+V

Aux+V

Mod+V

Impf

ANNE A1 2;3 A2 2;6 A3 2;8 A4 2;10 A5 2;11 A6 3;1 A7 3;3

– 2 6 4 7 7 4

– – + + + + +

– – – 1 1 1 –

– – – – – – 1

1 1 1 1 – – –

– 20 23 24 37 58 65

– 1 – 4 10 15 5

1 3 1 8 7

– – – – 1 – 1

JEAN J2 2;0 J3 2;2 J4 2;4 J5 2;6 J6 2;9 J7 2;11

– – 1 3 7 6

– – – – + +

– – – – 9 6

– – – – 5 9

12 6 1 1 – 4

1 1 8 6 92 74

– 1 – – 6 3

– – – 1 32 23

– – – – – 3

DANY D1 2;2 D2 2;6 D3 2;10 D4 3;2 D5 3;6

– – 1 26 8

– – + + +

– – – 3 3

– – – 10 8

3 2 – – 2

1 15 72 96 71

– 10 2 9 19

– (1) 33 26 32

– – – 5 1

MIMI M1 2;0 M2 2;2 M3 2;6 M4 2;10 M5 3;2

– 11 6 7 6

– + + + +

– – 3 2 2

– 3 1 4 4

– 3 – – –

13 58 67 83 104

– 7 15 11 10

3 6 12 7 11

– – – – 6

–

Legend: fV+Ng = finite Verb + Negation (pas); case = je/moi etc. in opposition; ocl = object clitic, in preverbal position; C = Complementizer, i.e. subordination with subjunctions; Nonf = infinitive- or participle-like forms, without auxiliary or preposition (here = root infinitives); Scl+V = subject clitic + verb; Aux+V = auxiliary (a/est etc.) + Verb; Mod+V = Modal verb (including vais/va etc.) + Verb; Impf = Imparfait. (There are about 200 utterances in each recording.)





Suzanne Schlyter

. Later development – subordination etc.: COMP is instantiated Later still, after the appearance of object clitics, the children also start using Imparfait (A5 2;11, J7 2;11, D4 3;2 and M5 3;2). In the first occurrences only the forms étais/t, avais/t and some modals (voulais/t etc.) are used. (12) oui, y avait là ‘yes, there was there’

(Anne 5, 2;11)

Unlike the adult learners, the children do not use present tense form to refer to past time (cf. Schlyter 1995). More or less simultaneously with Imparfait, the explicit marking of subordinations appears. Both are, however, rather rudimentarily used in the beginning, and a more productive and certain use appears after the ages shown here. (13) a.

on bougeait pas one move-impf not ‘we did not move’ b. regarde comme je fais! ‘look how I’m doing’

(Jean 7, 2;11)

This late appearance of subordinations is clearly different from the adult learners, who use such very early, practically as soon as they have other indices of INFL. For a general overview of the four children in these respects, see Table 1.

. Summary of the development in (2)L1 These children show development through the following stages: a.

A first stage without any auxiliaries, modals or subject clitics, nor a functional opposition between the same verb in different forms, i.e. no morphologic evidence for access to INFL. There are no postverbal negations, no case distinctions and no object clitics, i.e. no syntactic evidence for access to IP/INFL. b. A second stage (from 2;6 in Anne, Jean and Dany, and 2;0 in Mimi) in which subject clitics become productive and varied, auxiliaries and modals appear, and verbs start to have different forms that are functionally opposed. There are practically no errors in verb morphology. Postverbal negation appears, and somewhat later, object clitics. Case oppositions are marked. This constitutes morphologic as well as syntactic evidence for IP/INFL. c. A third stage (from about 3 years in most children) in which subordinations of different kinds, marked with subjunctions, start to appear. There is, thus, evidence for CP/COMP. More or less around this time, the children also start using Imparfait.

Development of verb morphology and finiteness

The transition between stage (a) and (b) shows that there is a point in time at which finiteness, in both a syntactic and a morphologic sense, appears, and that morphologic development is strongly related to syntactic development. The development also indicates that lower elements of the syntactic structure are used before higher ones, and the facts are compatible with a structure-building model or a truncation model. These findings support those of several earlier studies of French L1 development (Meisel 1994; Pierce 1992 etc.). It is evident that there is a later refinement of the functional and discursive use of these forms, which will, however, not be studied here.

. Adult second language acquisition: Early stages As for second language acquisition, it is well known that adult L2 learners have great problems with their verb morphology. An important problem is, as indicated above, how to account for the onset of verb morphology, or to establish criteria according to which we can say that finiteness is acquired. Let us, therefore, first discuss adult learners’ use of verb morphology according to more or less strict criteria, in the two least advanced L2 learners, Henry and Lisa.

. Finiteness in early stages according to restrictive and non-restrictive criteria If we use rather restrictive quantitative criteria for finiteness (cf. above), which take into account the general pattern or the majority of the cases and not just a few instances, then the less advanced learner, Henry, can be said to have not yet acquired verb morphology and to not yet mark finiteness. Henry uses a great number of utterances without any verb. These constitute 30 out of his 88 utterances, and they are not target-like elliptical responses to a preceding question, but are mostly entire predications initiated by himself. Example: (14) taxi dans la bois, taxi in(to) the wood après, pluie pluie pluie; then, rain rain rain; trop trop pluie, trop trop pluie. too too much rain, too too much rain. taxi à la train, dans la sud. taxi to the train, in the south.

(Henry)





Suzanne Schlyter

(He wants to say, roughly: “nous avons pris un taxi pour aller dans la forêt, mais après, il pleuvait tout le temps. Il y avait beaucoup trop de pluie. Nous avons pris un taxi pour aller à la gare, et ensuite nous avons pris un train pour aller dans le sud de la France.”) This pattern looks like what in the ESF project is called the “Pre-basic variety”, or the NUO (Nominal Utterance Organization), where the learner expresses mainly the nominal elements in focus (see Klein and Perdue 1997). It looks like a kind of structure truncated below IP, but is not completely without grammatical morphemes, since he uses prepositions and articles. As for the negations, the majority of these (21 out of 29) are of a very elementary kind, i.e. not postverbal as in the target language but occurring in the structure NEG+X – which is, according to many authors (Klein 1986 etc.), generally the first stage of Negation in L2 acquisition. (15) a.

la chambre, non grand lit the room, no big bed ‘in the room, there was no big bed’ b. je, personnellement, no chambre I, personnally, no room ‘I, for my part, had no room’ c. nepa de hôtel Paris, non non! neg of hotel Paris, no no ‘there is no hotel (room) in Paris, no no’

(Henry)

Among the 62 verbs he used, there were two constructions that can be considered as chunks (i.e., unanalysed units): c’est (23 occ) and j’ai (10 occ). The other verbs (19 types) seem to occur in one invariable form independent of their temporal or finite function (except the chunk ‘comprends pas’ contrasting with “comprendre”): (16) a.

je comprendre, la dame comprendre I understand-inf, the lady understand-inf ‘I (normally) understand and the lady (normally) understands’ b. le concierge comprendre, je comprendre the concierge understand-inf, I understand-inf ‘the c. understood and I understood’ (Henry)

His subject pronouns cannot be considered as instances of morphology, since it is clear that they do not have the status of clitics (like in the children). In the preceding example (16a, b), he accentuates JE as if it were a strong pronoun, contrasting it with a nominal such as ‘la dame’. Except for the chunk j’ai, he does not mark reference to the past. In 25 cases of past reference, he marks past tense 9 times (with j’ai + an invariable non-finite

Development of verb morphology and finiteness

form) and uses the invariable form 16 times. See example (16b) above and also (17), which refers to a past event: (17) Je traduis allemagne – française. I translate-prs German – French ‘I translated between German and French’

(Henry)

According to such an analysis, Henry is essentially at the basic or pre-basic stage, and cannot be said to have acquired verb morphology, or to have access to finiteness in the sense of INFL. However, if we use a less restrictive definition of the acquisition of verb morphology or of INFL, we can argue that he already has acquired morphology – even if he does not yet know every form – and that he has access to IP/INFL. According to these criteria, it is sufficient that the learner produces some single finite morphemes for finiteness to be considered as acquired. If so, his chunks j’ai and c’est can be considered as instantiations of the Functional Category IP, since they hold that place, even if the learner does not know enough lexemes to vary the position (Hawkins 2001: Chapter 2): (18) a.

J’ai contrôlé le système taxi ‘I have controlled the taxi system’ b. Je, j’ai appris . . . ‘I, I have learned’

(Henry)

According to such quantitatively non-restrictive criteria, we can also consider comprends pas as a productive case of postverbal negation, and not as part of a chunk, since Henry also uses negation with another verb: (19) a.

c.

je comprends pas I understand neg ‘I do not understand’ la concierge euh n connaît pas allemagne the concierge knows neg Germany ‘the c. did not speak German’

(Henry)

The opposition he exhibits between je and vous suggests that Henry does, at least partly, master subject-verb agreement. This is reinforced by one case in which he apparently tries to change the form of the verb after vous: (20) a.

je parle français très bon ‘I speak french very good’ b. vous comprendr-E français, non? you-2pl understand-2pl French, no ‘you understand French, don’t you?’

(Henry)





Suzanne Schlyter

Another learner, Lisa, produces a similar system of verb morphology at a somewhat more advanced stage than Henry. The first recording will be studied here. Although she has studied French for six years at school, and received good grades in French, she has not clearly acquired a productive use of French verb morphology, and her spontaneous oral production is rather similar to that of the Basic Variety in the ESF learners (cf. Klein and Perdue 1997). Like Henry, she has problems distinguishing the finite (short) form from the infinite (long) form (in the gloss referred to as NF): (21) Lisa:

Int: Lisa:

il change, he change-prs ‘he changes his mind’ et il décidE à monter # encore he decide-nf to go up again ou comment on dit? montrer? monter? encore? or how one says? montrer? monter? again? dans sa chambre. into his room et de rester. and to stay-nf il restE. he stay-nf ‘he stays’

(Lisa 1)

As we can see from (21) and from (16), adults use subjects, mostly subject pronouns, with the infinite forms (see further Table 2, page 35, “Psuj+V”) – a pattern that does not occur in the child data. Lisa also has problems in marking the past tense (ex. (22), last line), and her negations are predominantly of the type Neg-X: (22) Int:

Lisa: Int: Lisa:

Vous avez fait plutôt des traductions suédois-français ou plutôt la communication parlée? ‘did you rather do translations Swedish–French or rather spoken communication?’ NON beaucoup parlé (. . . ) no much spoken vous n’avez pas beaucoup parlé? ‘you did not speak very much?’ non, non pas beaucoup parlé. no, not much spoken

Development of verb morphology and finiteness

eh, crire et de(s) excercices (. . . ), eh, write and exercises ‘we were writing and doing exercises’ c’est imparfait de passé composé et # choses comme ça it is-prs imparfait of passé composé and things like that’ ‘it was Imparfait and Passé Composé and things like that’ (Lisa 1) This use of present (which occurs frequently) might give the impression that she lacks finiteness in form of past tense marking, but there is great variability (see Table 4 below, page 39). Example (22) continues with (23), illustrating that she sometimes uses a past tense marking: (23) et à la maison # at home, traduisE translate-?impf

on we les the

avait un # chapitre #, had-impf a chapter mots et comme ça. words and like that

(Lisa 1)

She does not always use the Neg+X type of negation, but also has some postverbal positions of the negation, often combined with a correctly used preverbal part of the negation. In c. the position is incorrect, after the non-finite verb: (24) a.

le end, je ne comprends pas the end, I neg understand neg b. je ne crois pas * le end I neg believe neg the end c. nous n’ a commencé pas we neg have started neg

(Lisa 1)

Like all the L2 learners studied here, there is a clear marking of subject-verb agreement at least with some (non-thematic) verbs: (25) a.

je suis (. . . ) vingt-deux (. . . ) I am-1ps twenty-two b. il est Français he is-3ps french

(Lisa 1)

This is different from the child data, in which such agreement patterns are avoided. (The reasons may be either that the L2 learners retain the entire chunk, or – in another framework – that the children use subject pronouns as affixes (cf. Granfeldt and Schlyter 2003), which makes further affixes redundant, whereas adult learners use subject pronouns as nominals, thus leaving space for a verbal suffix.5 )





Suzanne Schlyter

Lisa sometimes correctly uses Passé Composé for reference to the past, in the typical French “aorist” function (cf. Schlyter 1996), like in (26) where she also exhibits subject-verb agreement: (26) hier eh, on a vu #, j’ai vu Maigret yesterday one has-3ps seen, I have-1ps seen ‘M. (et) le clochard d dans le vidéo. and the vagabond’ on the video

(Lisa 1)

This indicates that, even if a speech sample has many instances of lacking Functional Categories like INFL, it seems extremely difficult to find a complete lack of them in the adult data. Even learners in the ESF project at the earliest stages, like Zahra and Abdelmalek in the very first recordings, do not completely lack such elements, but sometimes use c’est or jana, according to the available ESF database. This is clearly different from the production of the children, where we observe an initial period without any such elements.

. General view of the L2 learners’ early development Table 2 below shows quantitative data concerning possible access to IP in the early stages of the adult L2 learners. Data from only the least advanced learners are presented here. The table is established in the same way as for the children, i.e. according to isolated occurrences of syntactic phenomena and morphology. We can observe here that, differently from the children, these adult learners a.

produce some auxiliaries and some subjunctions (complementizers) even in the very early stages. Very soon, they also produce some Imparfait forms, and also modal+infinitive; b. have postverbal negations, from the first recording; c. they use a great number of subject pronouns, from the first recording; d. distinguish between the cases nominative (je, c’ etc.) and oblique (= strong forms like moi, toi etc.), from the first recording; e. the nonfinite forms in finite contexts do not decrease in number (cf. however, below) and do not disappear with the appearance of auxiliaries and modals, as they do in the children (cf. Table 1: J6, A5, D4 and M3); f. preverbal object pronouns appear before nonfinite forms have disappeared. We can further observe, from the data, that g. all learners use c’est abundantly (cf. also Bartning 1997); h. the learners often use subject-verb agreement between pronoun and verb (see examples (20), (25), (26)), at least with non-thematic verbs.

Development of verb morphology and finiteness

Table 2. Evidence for INFL and COMP in L2 learners, early stages, from syntax and morphology Learner

fV+Ng

Case

Ocl

C

Nonf

Henry 1–2

17

+

–

4

15

Martin 1 Martin 2 Martin 3

5 11 4

+ + +

– 1 2

5 7 10

Petra 1 Petra 2

18 30

+ +

3 4

Sara 1 Sara 2

15 12

+ +

Lisa 1 Lisa 2 Lisa 4

2 11 6

Sama 1 Sama 2 Sama 4

19 7 29

PsujV

Aux+V

Mod+V

Impf

93

10

–

7

3 7 8

144 156 165

1 20 31

– 16 24

– 3 9

4 11

7 5

205 294

3 39

20 15

1 1

4 1

10 9

6 2

277 154

6 5

10 9

2 1

+ + +

– 1 3

6 21 32

11 11 20

128 244 194

15 10 9

– 2 4

5 6 2

+ + +

– 3 6

6 16 52

2 10 11

125 98 158

6 7 16

11 4 6

2 9 2

Legend: fV+Ng = finite Verb + Negation (pas); case = je/moi etc. in opposition; ocl = object clitic, in preverbal position; C = Complementizer, i.e. subordination with subjunctions; Nonf = Nonfinite verbs, i.e. infinitive- or participle-like forms without auxiliary or preposition (= here normally in finite context); Psuj+V = pronominal subject + verb; Aux+V = auxiliary (a/est etc.) + Verb; Mod+V = Modal verb (including vais/va’etc.) + Verb; Impf = Imparfait. (There are around 300 learner utterances in each recording. Henry 1 and 2 are here considered as one recording, since data in each are few and he does not seem to advance.)

These observations can be interpreted as follows: –

– –

The use of auxiliaries, modals and complementizers (=a) indicates directly the presence of INFL and COMP; this is also valid for (g) since c’est can be seen as a placeholder for INFL (Hawkins 2001). The use of postverbal negation (=b) shows that there must be an INFL for the verb to raise to, in order to be placed above NEG. The early presence of subjects (=c) shows that there must be an IP for the subject to be placed in. Also the subject-verb-agreement (=h) indicates the presence of IP, since the pronominal (non-clitic) subject before a finite verb is placed in SpecIP (cf. Figure 1). Also the case oppositions (=d) indicate an access to IP, since only IP distributes nominative case (je) differently from nonnominative cases (moi, me).

The non-disappearance of the nonfinite forms (=e) is the only phenomenon that could suggest that IP is not accessible. However, such forms are still present even af-





Suzanne Schlyter

ter object pronouns have started to be produced preverbally (=f), which implies access to a Functional Category above VP (see Granfeldt and Schlyter 2003). Furthermore, the syntactic use of the nonfinite forms suggests that these verbs are perhaps raised to INFL, since we also find them – incorrectly – preceding the negation ‘pas’: (27) a.

je ne connaître pas; I neg know-inf neg (Lisa 2) b. mais le langue # ne fai f faire pas une système pour moi; but the language neg make-inf neg a system for me (Lisa 4)

Similar facts, concerning primarily the frequent use of subjects with non-finite verbs, have also been observed for adult L2 learners by Prevost and White (2000). In contrast, such structures never seem to occur in the children, after they have started to use subject clitics, auxiliaries, etc. The nonfinite verb forms also occur in subordinations, where a COMP is clearly present: (28) a.

quand on être dans la # la when one be-inf in the b. eh quand on voir français eh when one see-inf French

‘eller’ dans la pays ‘or’ in the country eh . . . eh

(Lisa 4)

(29) parce que si on utilisE le dictionnaire en même temps. . . because if one use-inf the dictionary at the same time ‘because if one uses (generic present) . . . ’ (Sama 4) Examples such as (28)–(29) indicate that there is no truncation below IP, i.e. if COMP/CP has been projected, then IP must also have been projected. To summarize, we can observe in these adult L2 data that, if we use the more restrictive criteria discussed here, or the purely syntactic criteria, nothing indicates that there is a phase at which INFL is lacking, as it seems to be in the children. This is the case even in the very early stages, after 3 months of residence (Sara) and with only passive knowledge of French (Henry). A study of the determiners and the noun phrases (Granfeldt 2000, 2003) showed a similar pattern: only the children have a stage without determiners, whereas all the adults studied here produce them from start. The results indicate that the adult L2 learners do have access to IP, i.e. finiteness in a syntactic sense, from very early on, but not to correct and functionally opposed verb morphology. This constitutes an argument in favor of the “Missing Surface Inflection Hypothesis”, proposed by Lardière (1998), and Prevost and White (2000). The problem is now, however, how to consider the nonfinite forms, i.e. the cases of lacking verb morphology. It is not the case that these forms are just optional,

Development of verb morphology and finiteness

subject to random choice and never acquired, but most of the learners do acquire this opposition at a very late stage, as we will study in detail below.

. Late development of verb morphology in the L2 learners As mentioned above, the children studied here do not use nonfinite forms in finite contexts (like after subject clitics) or vice versa, and they practically always mark reference to the past, instead of using present tense forms as the adult learners do (Schlyter 1995, 1996). For the L2 learners, this is quite different. However, the L2 learners do gradually acquire systematic use of morphology, in the sense of criteria D–E above. If we study the opposition between different forms of each single verb, we observe that Lisa, in the first recording, often uses most of the verbs (types) in one single form – or in random variation. She uses the verb “étudier” in one single form, /étudiE/. This form is used for present tense, for infinitive and after auxiliary (j’étudie vs. pour étudier / j’ai étudié). Examples: (30) a.

eh elle habite, elle # étudiE she lives, she study-nf b. j’ai étudiE à Umeå à I have study-nf at Umeå for

à Stockholm cette semestre in Stockholm this term trois années three years (Lisa 1)

The verb type parler is used in the forms /parl/ or /parlE/ – not in systematic opposition for finite vs. nonfinite, but at random. Among the verb types used with more than one occurrence, six verbs (types) are used in this way (étudier, donner, écouter, parler, manger, faire, lire), i.e. without a functional opposition, and four are used with a functional opposition (achète/acheter, commence/commencer, dit/dire, vais/aller). This gives the ratio 4/6 of +/– morphologically opposing verb types in the speech of Lisa 1 (see Table 3). The rest of the verbs occur in only one function, which means that we cannot know whether the learner knows how to use the forms in different functions, or whether she/he just uses the most frequent form as an invariant form. Some of these verbs are used correctly (aime, pleure, habite), and some are not (travaillE, décidE, restE), but these are not included in Table 3. For a similar view on verb morphology, see Pienemann (1998). A problem with the analysis of Prevost and White (2000) is that they do not consider the verb types and their possible opposition, or the fact that learners may use a single invariable form – a form that may sometimes look like a finite form and sometimes like nonfinite one; this method may possibly bias their results. In the ESF project (Klein, Dietrich, and Noyau 1995), on the other hand, invariable forms





Suzanne Schlyter

of a certain verb type or forms in random variation are considered as evidence of the learner not having acquired morphology. A pattern like the one shown here for Lisa is very normal in adult L2 learners; see further Table 3 below, for the learners studied here, and also similar data in Parodi (2000) and Prevost and White (2000). However, contrary to the claim of the latter authors, the learners studied here do not only use a default form (i.e. parlE) in all contexts, but often also a short form (= finite form) in nonfinite contexts, e.g. il a vient. For example, Sara 1 has 2 occurrences of nonfinite form in finite contexts and 4 finite forms in non-finite contexts. In any case, this varying pattern is strikingly different from that of the children in these and most other data. Setting up tables like this would, therefore, be of practically no use for child language, because of the lack of variation. There is, however, in most learners studied here, a development from very few verbs (types) marking the +/–finite opposition, to a clearly increasing number of such verbs. This development is clearly independent of their access to INFL, which, as we saw (in terms of the restrictive and partially syntactic criteria), occurs very early, possibly from the very beginning of their French speech. We show here the result of earlier studies also on the more advanced learners of the corpus (Karl, Johan, Knut, Ylva and Nina). In Table 3 (on +/–finite forms) we can observe such a development in most learners, in the sense that the number of verb types that occur in different, functionally opposed forms increases, whereas the number of those in invariable or random forms decreases. This can be observed in Lisa (though not very strikingly), partly in Sama, more clearly in Ylva, Martin, Petra and Karl. Johan hardly develops in this sense. The two learners who, according to most criteria, are the most advanced (they use object clitics, Imparfait etc.), Nina and Knut, are able to use bound verb morphology in a practically perfect way. (In spite of rich formal instruction on verb forms, there is no significant difference between the formal and the informal learners in this respect.) The time of this acquisition is somewhat varying. We can observe a clear improvement in the informal learners who have been in France for a long time: MarTable 3. Verb types with systematic opposition +/–finite forms vs. verb types without such an opposition, in L2 learners Lisa

Sama

Ylva

Nina

Martin

Petra

Johan

Karl

Knut

L1 4/6 L2 3/7 L3 2/2 L4 7/6 L5 4/3

S1 6/1 S2 2/3 S3 11/3 S4 7/7 S5 7/–

Y1 5/3 Y2 3/1 Y3 1/– Y5 5/–

N1 9/– N2 7/2 N3 6/– N4 6/1 N5 9/–

M1 1/2 M2 9/7 M3 8/4

P1 2/10 P2 7/7

J1 3/7 J2 5/7 J3 3/7 J4 7/7

K1 9/9 K2 8/3 K3 16/2

K1–2 12/2 K3 18/1

Development of verb morphology and finiteness

tin 3: 16 months, Karl 3: 14 months and Knut 1: 20 months. Thus, it seems that these learners need between one and two years of residence to acquire target-like verb morphology. As regards the marking of past reference, we mentioned that the children practically never omit the marking of reference to past, whereas this is standard for the adult learners. In early stages, there is a tendency for the adults to use an unmarked form (present or the nonfinite form) to refer to the past (see (16b), (17), (22)). Somewhat later, they increasingly use the Passé Composé as a general past marker, and also some Imparfait, above all in the form of étais/t and avais/t. The increasing correctness of the marking in this variable system is shown in Table 4 below. One could possibly object that tense marking is a pragmatic phenomenon, and that the learners do not have to mark past reference as long as it is clear from the context that they are speaking about past time. However, in that case it would be difficult to account for the very clear increase in the past tense marking, which seems to indicate greater mastery of these forms. As we can see from Table 4, there is a clear development in the learners Sama (2–5), Martin, and Karl, possibly in Lisa, and a high amount of marking in the advanced learners Nina and Knut. Here, the occurrences (tokens) are calculated, and not the types. (As before, there is no clear difference between the formal and the informal learners.) To summarize, the variability in the use of bound morphology (+/–finite forms in the different verb types) and of past tense marking (with auxiliary in most cases) is very strong in early stages, decreases over time, resulting in, for most of these L2 learners, a target-like morphological system after about two years or more.

Table 4. Verb tokens marking vs. non-marking past time reference in L2 learners of French Lisa

Sama

Nina

Martin

Petra

Karl

Knut

L1 25/30 L2 25/31 L3 24/52 L4 14/12 L5 31/30

S1 8/2 S2 15/17 S3 14/11 S4 16/8 S5 50/1

N1 18/– N2 33/9 N3 6/– N4 20/2 N5 55/6

M1 1/19 M2 15/10 M3 13/6

P1 3/1 P2 29/14

K1 13/24 K2 16/14 K3 100/8* K4 43/15 K5 10/–

Kn1 27/– Kn2 30/1 Kn3 14/–

Legend: numbers of verb tokens referring to past events marked for past / not marked for past (Ambiguous forms ending in -E are excluded. *Karl has here, furthermore, 29 of these ambiguous forms)





Suzanne Schlyter

. Conclusions and discussion We have seen above that the development of verb morphology in child (2)L1 and adult L2 learners, in spite of certain similarities, proceeds in a qualitatively different way. In the L1 learners, we can observe a first stage of complete lack of verb morphology or of other evidence for the instantiation of IP. When morphological evidence appears in the child data, syntactic evidence in the form of postverbal negation etc. also appears. A gradual development of the syntactic categories can be observed, defined here as VP, IP and CP, from the bottom up (in the sense of lexical instantiation of these categories). The children have a practically target-like morphology from the first time different verb forms are used. The data corroborate the view that for L1, morphological development precedes or interacts with syntactic development (Herschensohn 2000). We can assume, as many other studies have done, that in L1 development, morphology helps the child to get a grip on Functional Categories such as INFL, with subsequent use of syntax (see Clahsen, Eisenbeiss, and Penke 1996). In the adult L2 learners studied here, we have not been able to observe a developmental stage completely lacking evidence of access to Functional Categories like IP (or DP), at least if we use the less restrictive criteria that have often been used in this discussion during recent years (Hawkins 2001; Prevost and White 2000). Since the criteria used here are the same as for the children, i.e. the first occurrences, the data from the two groups are easily comparable. The adults all show early syntactic signs of finiteness in the sense of the category IP: immediate use of auxiliaries, subject pronouns, case markings, c’est, postverbal negations etc. Before verb morphology is well acquired, they use subordinations (=CP/COMP) and preverbal object pronouns, also signs of Functional Categories. However, their verb morphology is strongly deficient, variable and irregular, until very late in the development. It does, however, develop, and after about two years of stay in France or after one intense semester of university studies, it is fairly targetlike in the spontaneous speech of most of these learners. This means that one of the most important differences between child and adult language acquisition is that whereas children acquire morphology and syntax simultaneously, and perhaps use local cues like morphology to build up their syntactic structure, the adult learners already show signs of access to this structure – not only cognitively as is well known, but also in the more technical, syntactic sense used here. What they have to acquire is the exact and systematic use of the different word forms, i.e. morphology. Such a result seems to be quite compatible with the view of certain research based on a functional perspective (Hendricks 2000) showing that adults can use their already acquired discourse knowledge in acquiring a second language, but have to learn the specific forms of the target language.

Development of verb morphology and finiteness

However, the problem still remains as to whether it is from the very start of L2 acquisition that the Functional Categories are accessible (cf. the “Continuity Hypothesis” or the “Full Access” approach, Schwarz and Sprouse 1996), or whether the learner builds them up gradually, with the help of input in the form of morphological and grammatical evidence from the target language (cf. Hawkins 2001). An argument for the latter position would be the existence of a still earlier stage in adult L2 learners than the ones shown here, without any evidence of INFL. It is not impossible to interpret the data of Henry as showing instances of a transitional stage at which he is just starting to have access to INFL. But we have not been able to find, neither in these data nor in the available ESF L2 learner data of French, clear instances of adult L2 learner language without any indications of access to INFL, according to the least restrictive criteria. We do not know whether there might exist other learners at an even more basic level or whether it is not possible in any L2 data to find such patterns in their pure form. This is still an empirical question. Another problem for a structure building hypothesis in L2 acquisition is that, if it can be shown that there exists a very early stage without any indication of IP, and such a category appears after some time, then we must postulate two different points of acquisition: one for the syntactic category IP – very early, after about 3–5 months of residence, and another one for the acquisition of systematic verb morphology, occurring very much later, after one to two years of residence in the country. The facts seem easier to account for if we propose that IP (and possibly also CP), in its syntactic sense, is present and accessible to these L2 learners from the very start, and if we consider the acquisition of bound morphology as a more genuine case of acquisition. This acquisition might be – at least partly – dependent on factors traditionally used to account for L2 acquisition, such as input frequency, saliency etc. (Anderson 1993; Ellis 1994: 380–381), since it occurs late after much input or with intense training in school.

Notes * This research is supported by a grant from The Swedish Council for Research in the Humanities and Social Sciences (HSFR), grant number F0686/1998. Special thanks to Jonas Granfeldt, Petra Bernardini and also to the audience of the Euresco conference at Sant Feliu, October 2000, for valuable comments on a previous version. . Or one of its subcategories TP (= tense), AgrP (= agreement), possibly also AspP (= aspect) or even other categories. . Such were the criteria used by Schlyter (1997) in order to relate the appearance of object clitics before verbs to ‘finiteness’ or ‘acquisition of verb morphology’ in this restricted sense. . Or more precisely AgrS.





Suzanne Schlyter . We will not discuss here whether this is AgrOP, INFL or something else, only that it is a position to which such an element can be attached, by movement or as an agreement marker. . A more technical way would be to say that AgrS in children is instantiated by the clitic pronoun, whereas in adults it is instantiated by the verb form. The question needs further study.

References Anderson, R. (1993). Four operating principles and input distribution as explanations for underdeveloped and mature morphological systems. In K. Hyltenstam & Å.Viberg (Eds.), Progression and Regression in Language. Sociocultural, Neuropsychological and Linguistic Perspectives (pp. 309–339). Cambridge: Cambridge University Press. Auger, J. (1995). Les clitiques pronominaux en français parlé informel: une approche morphologique. Revue québécoise de linguistique, 24(1), 21–60. Bartning, I. (1997). Structuration des enonces et strategies referentielles a l’aide de la predication c’est X chez des apprenants avances et des locuteurs natifs. Travaux-deLinguistique Nov. 34, 65–90. Clahsen, H. (1982). Spracherwerb in der Kindheit. Tübingen: Narr. Clahsen, H. & Muysken, P. (1986). The availability of universal grammar to adult and child learners. Second Language Research, 5, 1–29. Clahsen, H., Penke, M., & Parodi, T. (1993/1994). Functional Categories in Early Child German. Language Acquisition, 3(4), 395–429. Clahsen, H., Eisenbeiss, S., & Penke, M. (1996). Lexical Learning in Early Syntactic Development. In H. Clahsen (Ed.), Generative Perspectives on Language Acquisition: Empirical Findings, Theoretical Considerations, and Crosslinguistic Comparisons (pp. 129–159). Amsterdam: Benjamins. Dimroth, C., Gretsch, P., Jordens, P., Perdue, C., & Starren, M. (this volume). Finiteness in Germanic languages: A stage-model for first and second language development. Ellis, R. (1994). The Study of Second Language Acquisition. Oxford: Blackwells. Ferdinand, A. (1996). The Development of Functional Categories: The acquisition of the subject in French. The Hague: Holland Academic Graphics. Giuliano, P. (2000). L’acquisition et l’expression des fonctions négatives en français et en anglais comme langues secondes: Confrontation d’Études Longitudinales et Apports Théoriques pour l’Acquisition en Milieu Naturel. Unpublished doctoral dissertation, Paris 8 University. Granfeldt, J. (2000). The acquisition of the Determiner Phrase in bilingual and second language French. Bilingualism: Language and Cognition, 3(3), 263–280. Granfeldt, J. & Schlyter, S. (2001). Acquisition of French subject pronouns in child and adult learners. Presentation at the conference Structure, Acquisition, and Change of Grammars: Phonological and Syntactic Aspects, Hamburg, 27–29 October 2000. To be published in the Acta from the conference in Working papers on Multilingualism from SFB Mehrsprachigkeit, Universität Hamburg.

Development of verb morphology and finiteness

Granfeldt, J. (2003). L’acquisition des categories fonctionnelles. Une comparaison du developpement du DP chez des enfants et des apprenants adultes. Doctoral dissertation, Institute of Romance Studies, Lund University. Granfeldt, J. & Schlyter, S. (forthcoming). Cliticisation in the acquisition of French as L1 and L2. Hawkins, R. (2001). Second language syntax: a generative introduction. Malden, MA: Blackwell. Herschensohn, J. (2000). The Second Time Around: Minimalism and L2 Acquisition. Amsterdam and Philadelphia: John Benjamins. Hendricks, H. (2000). The acquisition of topic marking in L1 Chinese and L1 and L2 French. Studies in Second Language Acquisition, 22, 369–397. Jansen, M.-L. (2000). Second language acquisition: from theory to data. Second Language Research, 16(1), 27–43. Klein, W. (1986). Second Language Acquisition. Cambridge: Cambridge University Press. Klein, W. (1998). Assertion and finiteness. In Dittmar et al. (Eds.), Issues in the theory of language acquisition (pp. 225–245). Bern: Lang. Klein, W., Dietrich, R., & Noyau, C. (1995). The Acquisition of Temporality in a Second Language. Amsterdam: Benjamins. Klein, W. & Perdue, C. (1997). The Basic Variety (Or: Couldn’t natural languages be much simpler?). Second Language Research, 13(4), 301–347. Lardière, D. (1998). Case and tense in the ‘fossilized’ steady state. Second Language Research, 14(1), 359–375. Meisel, J. M. (1991). Principles of Universal Grammar and strategies of language use: On some similarities and differences between first and second language acquisition. In L. Eubank (Ed.), Point Counterpoint: Universal Grammar in the Second Language (pp. 231–276). Amsterdam: John Benjamins. Meisel, J. M. (1994). Getting FAT: Finiteness, Agreement and Tense in Early Grammars. In J. M. Meisel (Ed.), Bilingual First Language Acquisition (pp. 89–129). Amsterdam: John Benjamins. Meisel, J. M. (1997). The acquisition of the syntax of negation in French and German: Contrasting first and second language development. Second Language Research, 13. Parodi, T. (1998). Der Erwerb funktionaler Kategorien im Deutschen. Tübingen: Narr. Parodi, T. (2000). Finiteness and verb placement in second language acquisition. Second Language Research, 16(4), 355–381. Pienemann, M. (1998). Processability Theory. Asterdam: John Benjamins. Pierce, A. E. (1992). Language Acquisition and Syntactic Theory: A comparative Analysis of French and English Child Grammars. Dordrecht, Boston and London: Kluwer. Prévost, P. & White, L. (1999). Accounting for morphological variation in second language acquisition: truncation or missing inflection? In M. A. Friedemann & L. Rizzi (Eds.) The Acquisition of Syntax (pp. 202–235). London: Longman. Prévost, P. & White, L. (2000). Missing Surface Inflection or Impairment in Second Language Acquisition? Evidence from Tense and Agreement. Second Language Research, 16(2), 103–133. Rizzi, L. (1993/1994). Some notes on linguistic theory and language development: The case of Root Infinitives. Language Acquisition, 3(4), 371–393.





Suzanne Schlyter

Schlyter, S. (1990). The acquisition of tense and aspect. In J. M. Meisel (Ed.), Two First Languages: Early Grammatical Development in Bilingual Children (pp. 87–121). Dordrecht: Foris. Schlyter, S. (1995). Formes verbales du passé dans des interactions en langue forte et en langue faible. In S. Schlyter (Ed.), L’acquisition bilingue simultanée pré-scolaire (=AILE 6) (pp. 129–152). Paris: Université de Paris VIII. Schlyter, S. (1996). Télicité, passé composé et types de discours dans l’acquisition du français langue étrangère. Revue française de linguistique appliquée, 1. Schlyter, S. (1997). Formes verbales et pronoms objets chez des apprenants adultes de français en milieur naturel. In Cl. Martinot (Ed.), Actes du Colloque International sur l’Acquisition de la Syntaxe en Langue Maternelle et en Langue Étrangère (pp. 273–293). Paris: Annales Littéraires de l’Université de Franche-Comté, 631. Schwarz, B. D. & Sprouse, R. (1996). L2 cognitive states and the Full Transfer/Full Access Model. Second Language Research, 12(1), 40–72. Wode, H. (1981). Learning a second language: An integrated view of language acquisition. Tübingen: Narr.

JB[v.20020404] Prn:24/06/2003; 14:02

F: SIB2602.tex / p.1 (35-104)

“Tinkering” with chunks Form-oriented strategies and idiosyncratic utterance patterns without functional implications in the IL of Turkish speaking children learning German Stefanie Haberzettl

.

Introduction

This paper aims to discuss a structural pattern found in the speech production of two Turkish speaking girls learning German. It combines the copula or auxiliary ist 1 and a lexical verb, mostly an infinitive, as in the following example: (1) Ein Junge ist die Fußball spielen. A boy is the football play-inf ‘A boy is playing (with the) football.’

(Me92 )

The infinitive is placed at the end of the utterance, which entails a serialization with the complement preceding the verb, simply spoken: OV. Ist is placed in the second position after a nominal constituent which can be qualified as subject or topic. This pattern is intriguing as it exists neither in German nor in Turkish. There are several reasons to look for a functional interpretation that has to do with the marking of temporality and/or aspect, which will be discussed in the second part of the paper after the description of the data base in Section 2. However, at the end of Section 3, it will turn out that the pattern with ist + lexical verb need not be associated with a specific temporal or aspectual function.3 Rather, the ist pattern is the output of the learners’ “tinkering” with salient chunks of the target language (TL) surface structures. As it will be shown in Section 4, these chunks are brought together in the idiosyncratic complex ist pattern, which serves as a purely formal precursor. From this precursor, the learners then easily bootstrap to complex structures that comply with formal and functional TL regularities. Thus my final conclusion will be that form follows form and function follows

JB[v.20020404] Prn:24/06/2003; 14:02



F: SIB2602.tex / p.2 (104-185)

Stefanie Haberzettl

form, in support of my general hypothesis on the L2 acquisition by children: Form prevails over function; the acquisition of a second language is in large part based on an orientation towards formal default structures. In the early stages of the acquisition process, L2 learners first and foremost rely on certain syntactic and prosodic surface structures of the L2, or parts of structures, which are learned and automatized as such, because they are frequent and/or salient. Consequently, learners may produce patterns or parts of patterns without making use of their full functional potential. I will argue that this is the case with the ist + lexical verb construction. Although this definitely is not a frequent or salient structural pattern of the target language input, but an invention by the learners, I hope to show that this contradiction can be resolved.

. The learners and the data base The ist + infinitive pattern came to my attention when I was studying the acquistion of verb placement in the L2 German by Ne and Me, two Turkish schoolchildren (Haberzettl 2000). At the beginning of the data collection, both girls were six years old. Although the first audio-tape recordings took place six months after their arrival in Germany, they were true beginners. A more than sporadic contact with German started only when they entered school. During the school term, the children were audio-taped on a monthly basis. Because of the holidays, the longitudinal data show some gaps. However, since the acquisition process in general can be qualified as rather slow, gaps of two months are not that problematic. The girls’ slow learning progress is due to the fact that they did not have much contact with the German language, since they attended a class with only Turkish children, a so-called bilingual class. From year to year, the portion of the lessons held in German by German teachers increased. These classes were ordinary school subject lessons, not special German classes. Especially during the first two years, however, the input situation must be characterized as very poor, also because of the girls’ restricted contact to German peers. Nevertheless, Ne and Me acquired German in an immersion setting, and this is also the reason why their acquisition of German can be qualified as natural or untutored.4 The audio-taped interviews (average length: 35 minutes) consist of free conversation and picture story telling. The picture story telling is sometimes rather free. In most cases the interviewer is asking quite general questions like “What’s happening then”? in order to elicitate further utterances. There are also special tests aiming at certain grammatical phenomena, which are not included here.

JB[v.20020404] Prn:24/06/2003; 14:02

F: SIB2602.tex / p.3 (185-211)

“Tinkering” with chunks

The following passage shall illustrate the corpus. It is part of an interview with Ne in her 9th month of L2 contact and shows an ist + infinitive utterance in its context. (2) I:5 Ne9:

I: Ne9:

I: Ne9:

I: Ne9:

I: Ne9:

I:

So schau, jetzt schaun wir uns mal die Bilder an, hm? ‘Look, now we’ll have a look at the pictures, hm?’ Ja ____ so. yes (pause) so ‘O.K.’ Wo ist denn das? ‘Where is this?’ Das ist ein, hm, das ist ein hm, mm _ Park. this is a, mhm, this is a mhm, mhm (pause) park ‘This is a park’ Hm? ‘Mhm?’ Das ist ein - . this is a ‘This is a - .’ Ein Park? ‘A park?’ Schwimmen. swim-inf (There is a swimming pool in the picture.) Ein Schwimmbad sagt man da, ja? ‘A swimming pool it is called, o.k.?’ Ja, Schwimmbad, das ist da Picknick machen. Yes, swimmingpool, this is there picknick do-inf (Children are having a picknick.) Mhm, wer macht denn Picknick? ‘Hm, who is having a picknick?’

Like in this passage dating from the beginning of the data collection, large parts of the early interviews consist not really of picture story telling, but of picture description. This changed rapidly when the learners were more advanced. The following passage is a small extract of a rather elaborated picture story telling: (3) Ne14: Ja, na, Katze auch, fangt die Quack, cat also catches the Quack (The main protagonist Quack, a little frog, is caught by a cat.)



JB[v.20020404] Prn:24/06/2003; 14:02



F: SIB2602.tex / p.4 (211-264)

Stefanie Haberzettl

und Quack geht im - - - er and Quack goes in the - - he (Quack is brought in jail.) Quack ist nicht Geld kommt [kommt = bekommt] Quack is not money gets (Quack has attempted a bank robbery, but he did not get the money.) und er geht - - im Polizei. and he goes to the police (Quack is brought in jail.) As can be seen in the last sequence, the use of lexical verbs is not restricted to the ist pattern, lexical verbs also occur without the “ist support”. However, during a certain period of time the ist pattern figures significantly in Ne’s L2 production. In contrast, Me produces it less frequently, though for a longer period than Ne. (In addition, she often employs another kind of ist pattern constructing ist with a past participle, mainly gemacht, ‘made’, as in Der Junge ist so gemacht, literally ‘The boy is so made’; I will come back to this in Section 3.2.) The following table lists Ne’s utterances (minimum of 35 per interview) that contain a lexical verb combined with at least one complement. (Besides that, there are many more instances of utterances with a lexical verb, but simple sequences like Kind ist schlafen – literally ‘child is sleep’ – are not included here.) It is necessary to add that, before her ninth months of language contact, Ne does not produce more than a handful of utterances containing a lexical verb. Table 1. Utterances containing a lexical verb V, a subject S or a subject ellipsis (S) and at least one complement X (n ≥ 35) (S) + X + V

S + ist + X + V

(S) + V + X

e.g.

Katze Maus essen. ‘Cat mouse eat’

Das Kind ist Auto spiele. ‘The child is car play’

Katze essen Maus. ‘Cat eat mouse’

Ne9 Ne10 Ne13 Ne14 Ne15

51% 20% 20% 3%

41% 6% 10% 9%

8% 74% 70% 88% 100%

JB[v.20020404] Prn:24/06/2003; 14:02

F: SIB2602.tex / p.5 (264-318)

“Tinkering” with chunks

. On functionalist interpretations The question arises whether the idiosyncratic pattern containing ist + lexical verb expresses a certain function in the domain of temporality or aspect. One could be inclined to assume so, due to the following reasons: Firstly, this pattern bears an association to the English progressive from, and perhaps it actually marks imperfective aspect (Tsit – time of situation – includes TT – topic time, cf. Klein 1994). Secondly, it is also similar to the German present perfect and may in fact mark perfect aspect (Tsit before TT).6 Thirdly, Ne and Me are not the only learners creating idiosyncratic structures, eventually in order to express quite subtle functions. Huebner (1989), Starren (1996) and Starren and van Hout (1997) report on similar structures found in the data of a Punjabi speaker of L2 English and of Turkish and Maroccan learners of L2 Dutch, which they claim to be interpretable in functionalistic terms.

. Ist as a marker of imperfective aspect? As we know, German does not have a progressive form in order to mark imperfectivity, it is not an aspect-prominent, but a tense-prominent language. In Turkish, tense-marking is grammaticalized as well, but since aspect must always be marked, Turkish can be qualified as aspect-prominent. Since in Turkish, aspect is marked synthetically by a suffix, the ist pattern can not be a calque from Turkish. However, there is a suffix -iyor bearing a progressive meaning, so that the Turkish learners know about a progressive aspect.7 (4) ˇcoˇJuk ev-e gid-iyor child house-dat go-prog ‘The child is going home (to the house)’ If something like a language-dependent thinking for speaking really does exist, the Turkish learners may wish to mark progressive aspect in their L2 German as well and try to do so. Slobin (cf. Slobin 1987; Berman & Slobin 1994) has coined the expression “thinking for speaking” for an L1 induced inclination to conceptualize an event, an action or a state in the real world in a certain manner which depends on the functions grammaticalized in the language in question. Learners with an aspect-dominant L1 like Turkish then might always be extensively aware of aspectual qualities in their L2 production as well, even if aspect marking is not grammaticalized in that language. They might be so “eager” to express aspect that they do not hesitate to do so in a creative manner, by inventing idiosyncratic means specially designed for that purpose – e.g. a certain kind of predication structure, where the verb ist is combined with a VP, the purpose of which might be to express



JB[v.20020404] Prn:24/06/2003; 14:02



F: SIB2602.tex / p.6 (318-382)

Stefanie Haberzettl

a protagonist being in the state of doing something, so that the beginning or the end of the action is not in the center of attention. A brief overview of the use or rather the non-use of a copula like German ist in Turkish will help to understand the following reflections (and will also be relevant for Section 4). Turkish has a marker expressing existence or presence, as in: (5) Oda-nIn orta-sIn-da bir kedi var / yok room-gen middle-poss3-loc a cat exist / exist = neg ‘There is a/no cat in the middle of the room.’ This existential predicate may be qualified as verbal, it takes the same suffixes as other verbs, for example -di in order to mark past tense. (6) Su var(-dI) water exist(-past) ‘There is (was) water.’ In German, existence can be expressed by the verb sein, ‘to be’, or geben, ‘to give’: Es gibt X – literally ‘It gives X’ = ‘There is X.’ In contrast to the German copular verb sein, ‘to be’, though, the Turkish language does not use a copula in order to express a simple predication, as in the following examples: (7) Hasan hasta Hasan ill ‘Hasan is ill.’ (8) Bu araba Murat-In this car Murat-gen ‘This is Murat’s car.’ (9) Biz-im ev eski 1pl-gen house old ‘Our house is old.’ Due to the contrast between Turkish and German, for Turkish learners the German copula ist in timeless copular contructions is a salient element of the TL. If they associate ist with non-resultativeness or non-telicity, ist may seem useful to them as an appropriate progressive marker. This reflection on the question whether and why the learners use ist to form an idiosyncratic progressive may sound rather speculative. Furthermore, independently of the choice of means, the mere consideration that learners should invent some structure in order to express something they need not express (in the L2) may seem already speculative enough. What we have to keep in mind though, is that learner varieties are more than a deficient version of the target system, and

JB[v.20020404] Prn:24/06/2003; 14:02

F: SIB2602.tex / p.7 (382-448)

“Tinkering” with chunks

that learners seem to be very inventive in using TL material – copular verbs among others – for purposes that are not necessarily TL-adequate. Thus, Huebner (1989) reports on one learner (R.) from the ESF corpus (cf. Perdue 1993), a 21-year old Punjabi speaking learner of English, who “develops a system for encoding viewpoint which, though internally consistent, reflects structures in neither his first nor the target language” (Huebner 1989: 111).8 Huebner hypothesizes that “this system reflects transfer at the functional level”, a phenomenon which should not be restricted to learners with L1 Punjabi: “Since all languages have ways of incorporating viewpoint, adult learners, it seems, would want to find ways of indicating it in their learner varietes as well, and from early on” (ibid.: 116). Interestingly, one of these ways to indicate viewpoint respectively to mark reference to participants is R.’s employment of is. R.’s interlanguage shows essentially three surface order utterance patterns (ibid.: 123): A. (is) NP1 (is) V NP2 (PP) B. (is) NP1 (is) V (participle / PP (PP)) C. is Adj / NP / PP

for transitives for intransitives for predications

There are several combinations of is +NP:9 Is occurs pre- and post-NP and even both (is + NP + is). What is important here is that NPs with is are usually restricted to the NP1 slot, the slot which can be associated with the status of a topic or theme, in complementary distribution to the + noun or one/quantifier + noun in the NP2 slot (the default case, 0 + noun, being found in either NP1 or NP2). NP + is is used for the first mention if an entity definite [sic!] by associative reference and to reintroduce referents [. . . ] after several utterances in which no reference is made to that entity or when there are two or more persons in the scene. In this sense, its function is similar to that of pronoun copy in the TL. (ibid.: 125)

The other linearization, is + NP, can carry the same function, but it can also encode a possessive relationship, or deictic reference in a focusing context (which is something completely different!). Compare the following set of examples including Huebner’s interpretations (cf. ibid.: 126): (10) and y’know girl [pause for 1 second] not working / is dad dead (= And you know the girl [= NP2 referential definite]? The one who was unemployed? The one whose father died?) (11) cake man is again to police ‘is not pinch is GIRL pinch’ is charlie again nothing eat (= The baker [= NP1 reintroduced 0 + N10 ] went back to the police-



JB[v.20020404] Prn:24/06/2003; 14:02



F: SIB2602.tex / p.8 (448-519)

Stefanie Haberzettl

man [= NP2 referential definite: 0 + N]. ‘He [= is: Charlie] didn’t pinch it. It was the girl [NP1 focused: is + N] who pinched it.’ There’s Charlie [= reintroduced: is + N] again with nothing to eat.) The list of different functions of is in combination with a noun is not complete yet. Huebner finds instances of is + NP introducing referents of three different kinds. First, a new, indefinite referent in NP1, secondly, a referent known to the listener of R.’s retelling of the story, but unknown to the protagonist whose perspective is taken, thirdly, a referent known to both the listener and the protagonist whose perspective is taken, but to whom the reappearence of the referent is unexpected. Cf. the examples (ibid.: 126): (12) and shop man is coming (not analyzed; remark by S.H.) is lady, looking (= The baker comes out. There’s a lady who sees all) (13) and is girl coming as well (which is, from the protagonist’s perspective, unexpected; remark by S.H.) Even though there are still several functions of is left, it is possible to interrupt the presentation of Huebner’s findings here, since the emerging picture is quite confusing already. The interpretations attributed to is are so opposing that one cannot help thinking of post-hoc explications for phenomena which should be explained otherwise. One cannot deny that in Huebner’s collection of examples, there are several instances of is belonging to one category, namely is as a device for focusing, but this device need not be regarded as a structure created by the learner which “reflects structures in neither his first nor the target language” (Huebner 1989: 111, cf. the complete citation above). Quite the reverse, learner utterances like the example above, is GIRL pinch (and not Charlie. . . ), do reflect the TL’s cleft construction It is X who. . . (and not Y), in syntactic and prosodic form as well as in function. The same holds for utterances like is lady, looking (see above), where a new protagonist (new to both the listener and the main protagonist whose perspective is taken) is introduced. Huebner himself proposes the paraphrase with the presentational there is . . . , who. . . – why not assume that the learner is aiming at this TL construction? The only is constructions in R.’s interlanguage which could not be considered a more or less successful imitation of the TL are represented by utterances like the following examples, where the function of is is claimed to be similar to that of a pronoun: (14) is (= Charlie) looking bill is (= bill) too much (0 = Charlie) got no money’

JB[v.20020404] Prn:24/06/2003; 14:02

F: SIB2602.tex / p.9 (519-574)

“Tinkering” with chunks

In this sample, Huebner detects “anaphoric reference to Charlie with is”, “deictic reference to the bill with is”, and “reference to Charlie again, this time with zero” (ibid.: 122). One could object, though, that this interpretation of is as an anaphor is not convincing (neither is it in the other examples, which cannot all be treated here in detail.) As for the first utterance in (14), why not assume that the learner intends to produce a progressive form, and simply omits the (thematic) subject, and in the second utterance, that he intends to produce a predication with the copular verb, again without expressing the thematical subject? And if we attribute verbal status to is, we need not be startled by the fact that in the third utterance, there is no is, because we would not expect a progressive of the lexical verb to have got. Not all instances of is in R.’s interlanguage can be associated with TL constructions that easily, since is is not only combined with predicative nouns/adjectives or -ing forms of main verbs, but also with participles (e.g. girl is said ‘you coming as well’ [ibid.: 122]), or with stems (e.g. Is cakeman coming. Is look. Another woman looking to that girl [ibid.: 130]). However, in all cases, interpretating is as a verbal element appears more appealing than treating it as a reference marker. Consequently, Huebners’s data neither corroborate the general hypothesis that learners sometimes use TL material in order to express certain functions in a non-targetlike way at all, nor do they give cause to analogically analyze ist in the ist + infinitive pattern found in the German L2 data as a contribution to a system of indicating aspectual viewpoint. In this respect, the findings of Starren (1996) and Starren and van Hout (1997) in the L2 productions of Turkish and Moroccan Arabic speaking learners of Dutch, and of Moroccan Arabic speaking learners of French (taken from the ESF corpus, cf. Perdue 1993) are much more convincing. These learners “seem to create a protoverbal morpho-syntactic system of tense and aspect regardless of target language specific properties” (Starren & van Hout 1997: 456), namely structures with two positions for “auxiliaries” (is among them) as distinct markers for tense and aspect, e.g. (ibid.: 460): (15) die was bij charlie is gestaan that (= the girl) was with charlie is stood The first auxiliary, was, encodes the topic time (TT), the past, for which the assertion is made. The second auxiliary, is, marks aspect, here imperfectivity. In the following example (ibid.), is encodes present tense, and heeft perfect. (16) dan is hij heeft werk aanvragen then is he has work ask On the one hand, these constructions seem complicated, but on the other hand, they can be considered as the result of a simplifying strategy. Firstly, the learners systematically split up tense and aspect, which are amalgamated in the TL in one



JB[v.20020404] Prn:24/06/2003; 14:02



F: SIB2602.tex / p.10 (574-625)

Stefanie Haberzettl

finite verb form, and secondly, they can rely only on free morphemes easy to cope with. Thirdly, scope relations are kept transparent, since the tense marker with scope over the whole utterance is placed in the “topic component of the utterance” (ibid.: 456), whereas the aspect marker introduces the VP, the locus of the event specification. In a way, this seems nearly too good to be true. However, the interpretations fit the data, and besides that, the double-auxiliary construction can be related to a predecessor pattern (cf. Starren & van Hout 1996) with two distinct positions for time adverbs as markers for topic time and for the time of the situation in relation to topic time, in other words: aspect. In the following example (Starren & van Hout 1997: 456) the first adverb places the event in time and the second adverb expresses iterativity:11 (17) vandaag ik altijd weg met auto today I always away with car The fact that there are structural parallels in the idiosyncratic positioning of time adverbs and auxiliaries cannot be taken as fully independent evidence for an interpretation of the latter in analogy to the former. However, one is more inclined to accept that there are idiosyncratic IL forms for certain functions if different observations can be conflated and applied to each other, and this seems to be the case here. And finally, even if the form-function correlation is not target-like and even if is, was and heeft could count as free morphemes in their own right, the learners take verbs from their TL input to express categories usually expressed by verbs. When the L2 speakers learn to “package these free morphemes in complex finite verbal forms expressing both tense and aspect” (Starren & van Hout 1997: 460f.), they will not have to give up the “material” (in contrast to the learner R. [Huebner 1989], who cannot keep the form is in his system of reference marking). The structures using two proto-verbal markers of tense/aspect are different from the ist pattern in the German L2 data, since the former include two auxiliaries, the latter only one. Nevertheless, they might have a common denominator. The Turkish learners studied by Starren (1996) and Starren and van Hout (1997) seem to use the auxiliary is in the L2 Dutch as a marker of imperfective aspect, and the interpretation of ist as a marker for imperfectivity fits in many of the utterances with ist + lexical verb in L2 German as well. However, the hypothesis that the ist + lexical verb pattern represents a kind of an idiosyncratic IL progressive form cannot be confirmed after all. Taking a second look at the examples, we realize that in (2), here partly cited in (18), the ist pattern is perfectly compatible with a progressive reading.

JB[v.20020404] Prn:24/06/2003; 14:02

F: SIB2602.tex / p.11 (625-688)

“Tinkering” with chunks

(18) Ne9: Ja, Schwimmbad, das ist da Picknick machen. Yes, swimmingpool, this is there picknick do (Children are having a picknick.) The problem arises that a progressive meaning can also be attributed to other utterances including a lexical verb, but no ist: (19) I:

Schau mal, was macht denn der Mann da? ‘Look, what is the man doing there?’ Ne9: Ein Eis. an icecream ‘Some icecream.’ I: Was macht der mit dem Eis? ‘What is he doing with the icecream?’ Ne9: Das Kind Eis kaufen. the child icecream buy ‘The child is buying some icecream.’ I: Mhm, er verkauft Eis. ‘Hm, he is selling some icecream.’ [to the child]

In example (3), here partly cited in (21), the utterance containing ist definitely does not have a progressive meaning: (21) Ne14: Quack ist nicht Geld kommt [kommt = bekommt] Quack is not money gets (Quack has attempted a bank robbery, but he did not get the money) Ist + lexical verb carries a resultative meaning here, and thus it is necessary to check if the learners systematically employ the ist pattern in order to express perfect aspect.

. Ist as a marker of perfect aspect? In fact, the ist pattern resembles the German present perfect, an analytic verb form consisting of an auxiliary in the verb second slot and a lexical participle in the verb last slot. In the target language, Quack ist nicht Geld kommt (21) would be (22). (22) Quack hat kein Geld bekommen. Quack has no money got ‘Quack has not got the money.’ Note that in German, the present perfect can not only be formed with the auxiliary haben, ‘to have’, but also with the auxiliary sein, ‘to be’.



JB[v.20020404] Prn:24/06/2003; 14:02



F: SIB2602.tex / p.12 (688-750)

Stefanie Haberzettl

(23) Er ist ins Zimmer gekommen. he is into the room come ‘He has come into the room.’ (24) Das Glas ist heruntergefallen. the glass is fallen down ‘The glass has fallen down.’ The German present perfect does not inevitably express perfect aspect (Tsit before TT), but it can also express the perfective (Tsit included in TT), which holds particularly for Southern German: Gestern habe ich Anna getroffen, literally ‘Yesterday have I Anna met (participle)’, is used synonymously with Gestern traf ich Anna, literally ‘Yesterday met (preterite) I Anna’ – “Yesterday I met Anna.” (Obviously, the aspect of a sentence in present perfect also depends on the lexical verb’s aktionsart or verb character.) It is not necessary to go into the details of the range of functions of the German present perfect (cf. Klein 2000), especially since the general question guiding this article is not how learners imitate the TL system, but whether they choose TL material to express certain functions defined in their interim grammar in certain stages of the acquisition process – e.g., perfect aspect or resultativity by means of the ist pattern. However, this is not an appropriate interpretation neither. It fits the example with the frog Quack, but not the other examples. Yet it does not follow that present perfect constructions in the TL input do not inspire the learner to create a similar construction. As a matter of fact, there may be an inspiration or a kind of reinforcement to reproduce a TL pattern – but eventually only on the formal level. In this context it is interesting to observe that the Turkish learners produce utterances – and do so in large numbers – which imitate TL present perfect constructions. Sometimes a perfective reading can be attributed to these utterances, but in most cases this would not be appropriate. (This holds for the very beginning of the acquisition process, in later stages the learners produce perfect constructions which are formally and functionally completely target-like; in other words: Function follows form.) (25) I:

Das is ein Zug, ja. ‘This is a train, yes.’ Me10: Ja, ein Zug, und ein kleines Katze. yes, a train, and a little cat I: Mhm. ‘O.K.’ Me10: Der is so gemacht. he is so done (The cat is holding on to the end of the train.)

JB[v.20020404] Prn:24/06/2003; 14:02

F: SIB2602.tex / p.13 (750-809)

“Tinkering” with chunks

Especially one of the learners, Me, produces the pattern X ist so gemacht, literally ‘X is so done’, extremely frequently for a couple of months. In one of the interviews, it is her most widely used pattern which allows her to speak fluently – without really saying anything. As already emphasized, this pattern does not express perfectivity, but like in (25), it often refers to an ongoing activity instead. This is illustrated by the following example, which is a little different since Me combines the participle of machen, ‘to make’, with the auxiliary to have in a target-like manner (the form habst itself is not correct, though). (26) Me10: Der mm die Jungen, die habst den ahm die Fest gemacht, ja. The hm the boys, they have the hm the party made (In the picture, children are having a birthday party.) I: Was für ein Fest hat er denn, wahrscheinlich? ‘What kind of party do they have, probably?’ These are just some examples taken from a huge amount of similar instances which allow for the conclusion that neither the ist pattern with a lexical infinitive nor the constructions similar to the German perfect with ist and a participle found in the first couple of months of the acquisition process can be explained as marking of either imperfective or perfect aspect.

. “Tinkering” with chunks Since the idiosyncratic ist pattern as well as the formally more or less targetcorresponding instances of a present perfect (which in fact express neither perfect nor perfective) are not the result of the learners’ desire to realize specific functional markings, it can be concluded that other mechanisms must be at work. In the following, I will propose a form-oriented explanation of the data. Admittedly, it is beyond doubt that language acquisition is driven by communicative needs, and consequently, by functions. It is also evident that, due to their L1 experience, L2 learners (even at the age of seven) already know about a whole range of specific, linguistically encodable functions, such as tense and aspect (in contrast to children acquiring their L1, who must learn about speech acts – beyond the expression of hunger or thirst – in general, cf. Tomasello 2000:64f.). Moreover, it is quite possible that the L1 knowledge causes “transfer on a functional level” (Huebner 1989: 111) – but, if so, this certainly happens in later stages of the acquisitional process. The first step inevitably consists of extracting linguistic material – entities and linearization patterns – from the input, and to reap as many benefits as possible of this material when confronted with the necessity to communicate. We observe the extreme case when a learner has at his disposal one, two or twenty pieces of formulaic speech



JB[v.20020404] Prn:24/06/2003; 14:02



F: SIB2602.tex / p.14 (809-851)

Stefanie Haberzettl

and when these pieces provide the only linguistic material for all his L2 speech acts, which obviously will thus entail inadequate utterances. The behavior of Me mentioned in Section 3.2 is an example for that strategy: Being confronted with the task of telling a picture story and lacking the necessary lexical material (especially verbs) to refer to the events in question, Me frequently falls back at the formula x ist x gemacht or x ist so gemacht (literally: ‘x is x made’, ‘x is so made’), instead of saying nothing at all. On the one hand, the comparison is adequate in that this strategy allows Me to speak fluently, on the other, the comparison is inappropriate, since normally, formulaic speak refers to a holistic imitation or reproduction of more or less complete TL utterances. Obviously, this is not the case here, given the erroneous combination of the auxiliary sein, ‘to be’, with the participle gemacht, ‘made’. As a matter of fact, it seems that Me has created a new pattern – a fixed serialization of elements including variables – by putting together two patterns or chunks from the input; just as the utterance schema x + ist (+ x) + lexical infinitive used by Me and Ne can be considered the result of such a tinkering with chunks. An (at least in some respect) similar phenomenon of “early syntactic creativity” is studied in an extensive documentation of a two-year-old child learning her L1 English by Tomasello et al. (forthcoming, reported in Tomasello 2000), who show that “the child was able to ‘cut and paste’ together her previously mastered constructions in order to create a novel utterance” (Tomasello 2000: 74). In another L1 study, Diessel and Tomasello (1999) looked at seven children’s first complex constructions including sentential complements and found “various kinds of ‘pastiches’ of various kinds of established item-based constructions” (ibid.: 76). The difference between these and the German L2 data lies in the fact that the chunks the L1 learners put together first existed as independent utterance schemes with a communicative function on their own, and that “in doing their cutting and pasting, children coordinate not just the linguistic forms involved but also the conventional communicative functions of these forms” (Tomasello 2000: 77). The tinkering strategy of the L2 learners is, as already emphasized, focused on the instantiation of formal patterns which are in the first instance not yet specified functionally. The chunks the L2 learners work with when forming the ist pattern are extracted from the input because these chunks are frequent and salient. First, there is the chunk XV, which largely prevails over the linearization VX (functionally, there is no difference between both). Table 1 shows that in the beginning of the learner’s L2 acquisition process, the XV chunk is found either separately, e.g. (27) Ne9: Ball spielen. ball play ‘Children are playing with a ball.’ or combined with a nominal constituent interpretable as a subject leading to S + X + V, e.g.

JB[v.20020404] Prn:24/06/2003; 14:02

F: SIB2602.tex / p.15 (851-910)

“Tinkering” with chunks

(28) Ne9: Das Kind äh Eis do, Eis, Eis kaufen. the child ehm icecream icecream, icecream buy ‘The child is buying some icecream.’ or combined with another chunk, namely x is, leading to the ist pattern, see below. German is an XV language, and although declarative sentences with a simple verb form show the serialization VX (er geht nach Hause, literally ‘he goes home’), there is much evidence in the input for the basic XV structure, due to instances of analytic constructions with auxiliaries or modal verbs, e.g. (30) Er ist nach.Hause gegangen he is home gone ‘He went home.’ (31) Er will nach.Hause gehen he want home go ‘He wants to go home.’ The fact that the Turkish learners already know left-branching VPs from their L1grammar may help them to detect the XV serialization in the L2 input, which in addition is prosodically prominent (cf. Haberzettl 2000: Chapter 3.2, on saliency and frequency of verb positions in the input). Or, in other words: If the Turkish learners start with an L1-induced OV hypothesis, they will find confirmation in the input. Thus, the learners start having at their disposal a pattern with a slot for the verb at the end: (I) NP-NP-V However, from very early on the learners realize that there is another position for a verbal element in the German sentence structure, due to very frequent instances of ist in the verb second slot – we can assume that learners are confronted with many utterances containing a copula in the input. For Turkish learners of German, this type of utterance must be quite conspicuous, since Turkish does not make extensive use of a copular verb such as sein, ‘to be’. However, Turkish employs the already mentioned existential predicate var: (32) Su var(-dI) water exist(-past) ‘There is (was) water.’ If we assume that Turkish learners frequently encounter TL patterns like (33) Das ist ein X – This is a X Da ist ein X – There is a X



JB[v.20020404] Prn:24/06/2003; 14:02



F: SIB2602.tex / p.16 (910-1004)

Stefanie Haberzettl

X ist ein Y – X is a Y X ist groß. – X is big. Wo ist X? – Where is X? and furthermore that learners are able to realize rather soon that ist is a verbal element (like the Turkish existential predicate var), we may conclude that they realize that there is a verb-second slot besides the verb-last slot in the German sentence structure. The learners produce utterances with the copula ist (and thus Pattern II below) even before they start to use lexical verbs. (II) NP-ist-NP The verbal element ist is tightly fixed to the verb-second slot, but it is of particular interest in this context that ist is not only used as a functional verb in the sense of copula or auxiliary, the latter being illustrated in (34) Me9: __Sonne ist weg. sun is away ‘The sun is gone.’ (35) Me10: Das ist ein Krakadil. [= Krokodil]. This is a crocodile. ‘This is a crocodile.’ In fact, there are many utterances where ist functions as a dummy for a lexical verb which the learner does not know yet or does not remember for the moment. (36) Me9: So, und die Junge ist die Boot. So, and the boy is the boat. (On the picture, the boy is manipulating the boat, he is not in the boat.) (37) Me10: Die Frau Fenberger ist ein Blatts. the Mrs. F. is a sheet of paper. (Mrs. F., the teacher, gave a sheet of paper to the children.) Nevertheless, Pattern II should not be translated into NP-V-NP yet, but at this stage should simply be considered a precursor of TL declarative verb-second sentences, because it is only the combination of Pattern I and Pattern II that can be considered a decisive step forward. In putting together the chunks X is and X V in a creative act resulting in the ist Pattern III, (III) NP-ist-NP-V

JB[v.20020404] Prn:24/06/2003; 14:02

F: SIB2602.tex / p.17 (1004-1045)

“Tinkering” with chunks

the learners realize in principle the complete German sentence structure with its two slots for verbal elements (left clause bracket and right clause bracket), cf. the learner utterances in (38) and Table 2: Table 2. A simplified overview of the German sentence structure initial field

middle field

verb, finite hat has kann can Hat Has conjunction dass That

Er He Er He

(38)

left bracket

das Bild the picture das Bild the picture er das Bild he the picture er das Bild he the picture

initial field left bracket middle field Ein Junge ist die Fußball a boy is the football

right bracket spielen. play

Das Kind the child

ist is

Auto car

spiele. play

Das this

ist is

da Picknick machen. there picknick make

right bracket verb, non-finite gemalt. painted malen. paint-inf gemalt? painted verb, finite malt paints

. Conclusion Due to the idiosyncratic ist pattern, which seems superfluously complicated at first sight, the Turkish learners will not have any problems with target-like verb placement in the following months of the acquisition process, neither with periphrastic verb forms in discontinuous position, nor with the verb final position in embedded sentences (cf. Haberzettl 2000). Thus we can conclude that form follows form and function follows form. Form follows form, because the target-like constructions with discontinous verb forms are hooked to a precursor, the ist Pattern III. And function follows form because in the beginning, the patterns with discontinuous verb forms (either ist + infinitive or ist + gemacht) do not cover other functions than utterances without ist, e.g. with respect to aspect marking, but represent pure formal variation. It is only in the later months of the acquisition process that the learners systematically associate different forms with different functions, namely



JB[v.20020404] Prn:24/06/2003; 14:02



F: SIB2602.tex / p.18 (1045-1105)

Stefanie Haberzettl

auxiliary in the left clause bracket + participle in the right clause bracket with perfect aspect, or verb last with subordination. The positive repercussions of the idiosyncratic patterns with ist are mentioned only briefly, because the issue of this paper is not what these patterns are good for in the ongoing acquisitional process, but whether their genesis has to be explained in functionalist terms. This does not seem to be the case. The data support the hypothesis that L2 learners first concentrate on detecting patterns in the input which are used as automatized routines, allowing for a fluent production of utterances. The idiosyncratic ist patterns with lexical infinitive or participle prove that the learners do not confine themselves to mere reproduction (and overgeneralization) of holistically extracted utterance schemes, but use this TL material creatively to construct new utterance schemes. Nevertheless, this tinkering with chunks still has to be taken as a form-oriented and tightly input-relying L2 production strategy.

Notes . In fact, one cannot decide in advance as to which category ist might belong. . Me9 = recording of the learner Me in her ninth month of L2 contact. . With specific I mean that it should have another temporal or aspectual reading than that of the other structural patterns found in the learners’ speech production in the same period of time. . See White and Genesee (1996: 142f.) who tested the linguistic behavior of different groups of L2 learners and concluded that a few hours of foreign language teaching are of no consequence if the learners otherwise live in an immersion setting. . I = interviewer . Note that several verb classes in German form the present perfect with the auxiliary sein, like in Er ist in Berlin angekommen – ‘he has arrived in Berlin’. . The Turkish sentences are taken from ErguvanlI (1984) and van Schaaik (1996). . Viewpoint refers to the “linguistic encoding of the physical orientation or perspective of the speaker to the events related, and to the participants in those events” (Huebner 1989: 116). . Huebner indicates that NP1 “is usually not lexicalized with a full noun phrase, but rather takes the form of zero, is, or in rare instances a pronoun” (ibid.: 123). Thus, utterances with double is should be possible, but they do not occur in Huebner’s examples, in contrast to serializations of is (interpreted as a kind of anaphor filled into the NP1 slot of A) + V, e.g. “is (= Charlie) looking bill” (ibid.: 122). I will come back to this in the main text. . Obviously, in “cakeman is”, Huebner does not interprete is as accompanying the noun, but as a verb.

JB[v.20020404] Prn:24/06/2003; 14:02

F: SIB2602.tex / p.19 (1105-1193)

“Tinkering” with chunks . Usually, iterativity is associated with the grammatical categorization of aktionsart or verbal character, and not aspect, but Starren and van Hout subsume iterativity under the category of “aspect notions” or “aspectual notions” (ibid.: 456, 459).

References Berman, R. & Slobin, D. I. (1994). Relating events in narrative: a crosslinguistic developmental study. Hillsdale / London: Erlbaum. Diessel, H. & Tomasello, M. (1999). Why complement clauses do not include a thatcomplementizer in early child language. Proceedings of the Twenty-Fifth Annual Meeting of the Berkeley Linguistic Society, USA, 86–97. ErguvanlI, E. E. (1984). The Function of Word Order in Turkish Grammar. Berkeley and Los Angeles: University of California Press. Haberzettl, S. (2000). Der Erwerb der Verbstellung in der Zweitsprache Deutsch durch Kinder mit typologisch verschiedenen Muttersprachen: Eine Auseinandersetzung mit Theorien zum Syntaxerwerb anhand von vier Fallstudien. Unpublished doctoral dissertation, Universität Potsdam. To appear in Linguistische Arbeiten. Tübingen: Niemeyer. Huebner, T. (1989). Establishing point of view: the development of coding mechanisms in a second language for the expression of cognitive and perceptual organization. Linguistics, 27, 111–143. Klein, W. (1994). Time in Language. London: Routledge. Klein, W. (2000). An Analysis of the German Perfekt. Language, 76(2), 358–382. Perdue, C. (Ed.). (1993). Adult language acquisition: Crosslinguistic perspectives. Volume 1: Field methods. Cambridge: Cambridge University Press. Schaaik, G. van (1996). Studies in Turkish Grammar. Wiesbaden: Harrasowitz. Slobin, D. I. (1987). Thinking for Speaking. Proceedings of the Thirteenth Annual Meeting of the Berkeley Linguistics Society, USA, 435–445. Starren, M. (1996). Temporal adverbials as a blocking factor in the grammaticalization process of L2 learners. In T. van der Avoird & M. Corsius (Eds.), Proceedings of the CLS Opening, Academic Year ’96–’97 (pp. 1–16). Tilburg: Center for Language Studies. Starren, M. & Hout, R. van (1996). Temporality in learner discourse. What temporal adverbials can and what they cannot express. Zeitschrift für Literaturwissenschaft und Linguistik, 104, 35–50. Starren, M. & Hout, R. van (1997). Do temporal adverbials shape morpho-syntactic tense and aspect marking? In A. Sorace, C. Heycock, & R. Shillcock (Eds.), Proceedings of the GALA’97. Conference on language acquisition (pp. 456–461). Tomasello, M. (2000). First steps towards a usage-based theory of language acquisition. Cognitive Linguistics, 11(1/2), 61–82. Tomasello, M., Lieven, E., Behrens H., & Forwergk, H. (forthcoming). Early syntactic creativity: A usage based approach. Submitted for publication. White, L. & Genesee, F. (1996). How native is near-native? The issue of ultimate attainment in adult second language acquisition. Second Language Research, 12(3), 233–265.



Finiteness in Germanic languages A stage-model for first and second language development* Christine Dimroth, Petra Gretsch, Peter Jordens, Clive Perdue, and Marianne Starren

.

Introduction

Finiteness is traditionally associated with the morpho-syntactic categories of person and tense. In this paper we distinguish, however, between the semantic concept of finiteness and the way languages mark it (for example with the help of verbal morpho-syntax as in standard German or Dutch).1 Following Klein (1998), semantic finiteness results from two separate pragmatic operations which we term here “anchoring” and “linking”. Anchoring provides the spatio-temporal co-ordinates into which the rest of the utterance is embedded (e.g. the topical place and time about which the utterance makes a claim), and linking validates the state of affairs expressed in the utterance for these coordinates.2 Linking expresses that this state of affairs is indeed true for the particular temporal-spatial anchorpoint talked about. In languages such as Dutch, German or English, both anchoring and linking are typically marked by one form, for example a finite copula as in the following example from Klein (1998: 226). (1) The book was on the table.

This fusion of two types of information in one form can be made explicit when contrastive stress (indicated by italics) is put on the copula as in (2). (2) The book was on the table.

Unlike full verbs, copulae und auxiliaries do not carry any lexical meaning proper. Contrastive stress therefore results in highlighting the meaning components of finiteness. It can provoke both a tense contrast, and a contrast regarding the claim.



Christine Dimroth, Petra Gretsch, Peter Jordens, Clive Perdue, and Marianne Starren

There is a tense contrast, when (2) is opposed to The book is on the table. What is highlighted here is the anchoring information, namely that the book’s being on the table occurred at some moment in the past. There is a contrast regarding the claim, when (2) is opposed to The book was not on the table. What is stressed is the linking function, which expresses that the book’s being on the table was indeed true for some spatio-temporal anchorpoint. Anchoring and linking contribute to the meaning of finiteness. Judging if some state of affairs did indeed hold true is only possible with respect to the relevant topic information about which the claim was made. While both functions are fused in one finite verb form in the target languages under investigation, child and adult language learners clearly separate the two. Initially, they express linking only, that is, the elements used for that purpose (e.g. particles and adverbials) do not adapt their form to the anchoring information. Example (3) from a second language learner of Dutch illustrates the point in case. Here wel (indeed) is used to indicate a particular claim, while anchoring with respect to topic time receives no explicit formal expression. (3) ik wél hard rijen I indeed fast drive

Only later in the acquisition process this is done with the adequate target language means of morphological finiteness. It allows the learner to express the spatiotemporal anchoring information, too. In what follows, we will therefore take the linking function of finiteness as a starting point and investigate which kinds of linking devices are used by different types of learners (namely children acquiring their L1 and adults acquiring an L2) in order to express the semantic concept of finiteness before target language adequate means for morphological finiteness marking on verbs are acquired. As we have seen in (3), the first linking expressions do not stem from the category of verbs. Lexical verbs are of course used in the part of the utterance that refers to the situation expressed (hard rijen in (3)), but they are not made finite in order to make them also suitable as linking-expressions. At that stage of acquisition, purely functional carriers of finiteness (e.g. auxiliaries) are absent, too. Learners avoid the complex verbal morphology (that not only involves markings relevant to the topic time, but also person and number agreement) and prefer lexical solutions when linking needs to be marked explicitly at all. It is only at later stages of acquisition that learners develop linking elements that can carry the specific combination of information relevant to the semantic concept of finiteness in one form, namely finite verbs. At the same time, learners must work on the devices for context embedding as well, since the form of the finite verb depends on the kind of (temporal) anchoring chosen.

Finiteness in Germanic languages

This means that, in order to express the semantic function of finiteness, any learner of a language has to come to master 1. the devices which allow the utterance to be appropriately embedded in the context; 2. the devices expressing the relation between the context-embedding and the state of affairs expressed in the utterance. The verb “master” then involves both understanding the concepts involved in these operations, and knowing how to use the relevant devices. The learning task is both conceptual and structural, and it is precisely its dual nature that makes the child learner-adult learner comparison interesting. Formal comparisons have been made between first and second language acquisition of (morphological) finiteness marking and related features: verb morphology, case assignment, subject-verb agreement (Meisel 1991, 1997; Schlyter, this volume). But there has never been, to our knowledge, any attempt to straddle the formal/functional split in this area, and in doing so, to evaluate the formal and cognitive complexity of the acquisitional task for both types of learners. Such is the aim of this study. It is in many ways tentative, and examines a very small range of languages. But the writing of it was thought-provoking, and we hope the reader will share our interest. In this chapter we compare the development of finiteness marking by children acquiring German and Dutch as their first language with the development of finiteness marking by adults acquiring Dutch and German as a second language. After giving the most basic information units (topic, state of affairs and the linking device itself), their interrelation in (simple) utterances and their relation to similar information units in prior utterances (Section 2), we will describe three acquisitional stages of finiteness marking (Section 3). At the first stage, the Holistic Stage, the linking between the state of affairs and the topic is achieved by pragmatic means, at the following Conceptual Ordering Stage this is achieved by lexical means, and at the Finite Linking Stage, by morpho-syntactic means. In conclusion, we will give a nuanced picture of both child and adult acquisition showing more similarities in both acquisition processes than is usually assumed.

. The conceptual structure: topic, predicate and linking In the introduction we have discussed the idea that, in order to be fully interpretable, information about any situation must be given with respect to some spatio-temporal anchor point. We shall call topic the expression(s) that refer(s) to the temporal and/or spatial and/or personal characteristics of the event that





Christine Dimroth, Petra Gretsch, Peter Jordens, Clive Perdue, and Marianne Starren

the utterance is about. The state of affairs that is claimed to hold for the topic is termed the predicate.3 Successful communication is dependent on the speaker’s and hearer’s mutual agreement on what is topical and what is predicative information. Linking elements express a more abstract grammatical operation. Linking devices do not carry content information, their function is rather to express that the content of the predicate is not just a mere idea but claimed to hold for the topic situation in question. By executing a linking operation the speaker takes on the utterance.4 There are different ways to express linking: it might be explicitly expressed by target language adequate (grammatical) means, or by intermediate (lexical) learner solutions. Furthermore, both, learner and target varieties permit, under certain conditions, the linking operation to be left implicit, or rendered by a special intonation pattern. The ‘conditions’ are however exceptional for the target languages, and more the rule for learner varieties.

. Topic: The anchor point The information unit called topic provides a (spatio-)temporal anchoring point for the information referred to in the predicate. The term anchoring is used here to cover the basic functions of topics, namely (a) the identification of what is talked about and (b) the embedding of the actual utterance in a discourse world. The discourse world can be text-internal (text, anaphor), or text-external (situation, deixis). While (a) enables the hearer to select the adequate piece of information about which something is said in the utterance, (b) makes sure that this piece of information is given the right information status with respect to the preceding discourse and the context. Within the restricted repertoire of a learner variety, these principles may sometimes get into conflict. While for example full NPs seem to be a secure means to fulfil topic condition (a) successfully, they might turn out to be inadequate or even misleading with respect to (b) in cases of reference maintenance (cf. Hendriks 2000). On the content level, adequately anchoring the utterance often implies the specification of a topic entity (e.g. a personal reference point) besides the (spatio-)temporal reference point. The form this information takes varies as a function of the speech act, of the situation, and of what has just been said. For the temporal context, a situation can be contextualised in relation to the here-and-now, or in relation to a previously mentioned situation. For the personal reference point, the topic entity can in principle be chosen from the verb-argument structure of the utterance. Even if there are prototypical topic entities (e.g. agents), every other element of the predicate information could be taken out to fill the slot for the topic entity in order to identify what is talked about. Agent often conflates with gram-

Finiteness in Germanic languages

matical subject, but the grammatical subject is also not an obligatory part of the topic information either. The topic may contain explicit reference to the temporal and the personal context co-ordinates; if not, they are inferable. There is, however, a crucial difference between the two items of topic information. In order to be interpretable, semantically finite sentences need a temporal reference point5 while a personal anchorpoint is not always present. This difference emerges clearly in the traditional distinction between (finite) categorical and thetic sentences. In the latter, reference to a topic entity is blocked by a non-referential form (there or it in English), but these sentences are always predicated on a pragmatically determined time span. The temporal reference point appears thus to be the minimal and obligatory topic element, even if its concrete shape must often be inferred from context.6 As a consequence, there are no topic-less sentences, and the temporal anchorpoint in the topic cannot be negated. At the basic stages of language acquisition (and beyond), topics are typically placed in initial position. The formal repertoire of topic expressions comprises zero anaphora, temporal and local adverbials, nouns and pronouns. The actual use of these depends on the learner’s (source and) target language and on the learner’s linguistic and cognitive development. As we will see in Section 3, L1 and L2 learners differ significantly with respect to the way topic information is expressed.

. Predicate: The (expression of the) state of affairs The predicate specifies some state of affairs that is claimed to hold for the topic. State of affairs is thereby used as a cover term for ‘situation’, ‘event’, ‘state’, ‘property’ and the like, but also includes simply referring to some entity as in thetic sentences. Its typical pragmatic role is to carry new or contrastive information, but depending on the information distribution and discourse embedding of the utterance, it may also contain maintained, non-contrastive information. The predicate seems to be the least constrained with respect to the content information given. In the learner varieties under investigation its prototypical position is to the right of the utterance. This is in accordance with the well known “focus last principle” (cf. Klein & Perdue 1997). The formal repertoire of predicative expressions comprises VPs, but also NPs (in presentational utterances) or adjectives (in attributive utterances). Early second language learners have been shown to have idiosyncratic lists of nouns and adjectives that are used as predicative expressions in the same way as verbal predicates. They can “bind” other constituents as arguments but do not occur as arguments of other predicative expressions (cf. Skiba 1993: 212).





Christine Dimroth, Petra Gretsch, Peter Jordens, Clive Perdue, and Marianne Starren

. Linking: The role of the validation device Linking is the complex of operations which allows the speaker to assert, question or request a state of affairs in relation to a topic. Grammatical finiteness (as one prominent possibility to express linking) is taken to be an (abstract) category that operates on an informationally-structured proposition. Its function is to validate the relation between the state of affairs described in the predication part of this proposition and its topic, which necessarily comprises a temporal reference point and perhaps other components, as we have seen. The most straightforward relation is a speaker’s assertion that some predicate holds for the topic. But linking can also take place with some different illocutionary force (it is possible to ask if the predicate is or is not true for the topic, or to request that the predicate be or be not true for the topic, etc.). Furthermore, the linking relation can be made more informative with the help of particles and adverbials: yes in addition to x; yes and not for the first time; yes perhaps, etc. As we shall see in Section 3 some of the elements that can semantically enrich the linking function of finiteness turn out to be its precursors in early stages of acquisition. Finite verbal morphology is the typical reflex in Dutch, English, and German of the linking operation. It is important to note, however, that even in fully fledged varieties of these languages, linking can be left implicit. Depending on context and intonation, merely uttering a predicate to someone can imply the linking operations of relating the predicate to a topic and asserting that / questioning if (etc.) this relation holds (cf. Lasser 1997). In what follows, we take an information structural approach and split utterances up into the three information units (topic, link, predicate) mentioned above. The topic of the relatively simply utterances we will be dealing with in the following often consists of reference to some time-span and/or to the external argument of the verb. The predicate expresses the state of affairs that is claimed to hold for the topic. Before morphological finiteness marking is acquired, this relation is established through lexical linking devices and we will now look at how the use of these linking devices develops over time.

. Acquisitional stages In the following we will show developmental evidence of the acquisition of finiteness as a grammatical linking device. The data originate from studies on the acquisition of Dutch and German by children learning these languages as their native language and adults learning these languages in an untutored second language environment. Most of the L1 and L2 data were collected longitudinally, the L1 data over a period as indicated in (4), and the L2 ESF-data and P-Moll data in three

Finiteness in Germanic languages

cycles of 10 months each. In the examples below a reference such as, for example, ‘Mahmut 1.2’ means that the utterance by Mahmut occurred at cycle 1, recording session 2. (4) Data sources. I = L1 Italian, MA = L1 Moroccan Arabic, P = L1 Polish, R = L1 Russian, T = L1 Turkish L1 Dutch

diary data

Jasmijn (1;7–2;2) Andrea (1;8–2;4) L1 German recorded data Benny (1;11–2;9) Lisa (2;0–2;8) Valle (1;11–2;1) L2 Dutch ESF audio recordings Mohamed (MA) Fatima (MA) Mahmut (T) Osman (T) Ergün (T) L2 German ESF audio recordings Cevdet (T) Angelina (I) Marcello (I) Tino (I) Gina (I) P-Moll audio recordings Janka (P) Urszula (P) Ivan (R) Wladimir (R)

As outlined in the previous section we will choose the perspective of information structuring for our analysis of utterance structure. This analysis provides insight into language learning as a developmentally staged process. Although stages of acquisition are part of a developmental process in which one stage is eventually given up in favour of another, in the actual process of language acquisition stages may overlap, i.e. properties of different stages may cooccur. For some time learner languages may even show evidence of backsliding when learners regress to a stage that they had seemed to have given up. Stages in language development are defined with respect to coherent systems of linguistic structure used by learners of a particular target language for at least a certain period of time. Our analysis of developmental progress within the four language learning situations mentioned above provides evidence allowing us to identify learner varieties at three levels of acquisition. In the following we will give a description of these three types of learner varieties used at consecutive stages of linguistic development. These developmental stages are termed the Holistic Stage, the Conceptual Ordering Stage and the Finite Linking Stage. A common property at each of these stages of acquisition is the ex-





Christine Dimroth, Petra Gretsch, Peter Jordens, Clive Perdue, and Marianne Starren

pression of a topic (explicitly or implicitly) and a state of affairs, such that the state of affairs is claimed to hold for the topic. This relation is established through what we called (§2.3) a validation or linking device. It is this relation of linking which receives different linguistic expression at consecutive stages of language acquisition. At the Holistic Stage validation is achieved by pragmatic means, at the Conceptual Ordering Stage by lexical means and at the Finite Linking Stage by morphosyntactic means. Interestingly, however, these validation devices are remarkably similar across the different L1 and L2 language contexts we investigated.

. The Holistic Stage: Pragmatic validation In early learner languages, frequently no validation device is expressed at all. We often find some topic and predicate in a juxtaposition relation and even predicates occurring alone. Predicates are commonly realized by a full noun, an adjective or a verb particle as well as by a verb. If modal operators are used, they are precursors in the acquisition of the functional properties of finiteness. Characteristic is the use of the negator nee (no) in L1 Dutch and nein (no) in L1 German. It functions as a modal operator with scope over the clause structure as a whole. In utterances with a topic and a predicate it typically precedes or follows both constituents as a unit. In early child Dutch nee (no) most prominently functions as a clausal operator with a negative modal meaning. Examples of Jasmijn’s earliest data at (1;9) are given in (5). (5) L1 Dutch. Holistic ‘nee’ (no) in Jasmijn [context]

‘nee’ Topic nee Peter nee Peter nee sepel

Predicate bij zitte, no P with sit, da zitte no P there sit (Jasmijn 1;9) emmer gooie dees no spoon bucket throw this (Jasmijn 1;9) nee menneng Cynthia hebbe no milk C have (Jasmijn 1;9) nee Peter fles ophale no Peter bottle get (Jasmijn 1;9) poessie bal pele. nee Cynthia afpakke kitty ball play. no C snatch away (Jasmijn 1;9) pop fles. nee Mijnie drinke doll bottle. no M drink (Jasmijn 1;9) nee pop ook valle no doll too fall (Jasmijn 1;9) Peter drinke. nee Mijnie drinke P drink. no M drink (Jasmijn 1;9) Peter nee poes tafel krimme P [says] no kitty table climb (Jasmijn 1;9)

Finiteness in Germanic languages

In Andrea (1;8–1;11), the use of the modal operator nee in clause-initial or clausefinal position is not that prominently present. As shown in (6), Andrea has a few instances of nee in clause-final position, only. (6) L1 Dutch. Holistic ‘nee’ (no) in Andrea [context] Topic Predicate ‘nee’ Mijnie toet nee M desert no (Andrea 1;9) oef eten nee dog eat no (Andrea 1;10) Mijnie in nee M in no (Andrea 1;10) papa maakt. pop pot nee daddy made. doll broken no (Andrea 1;10)

With respect to its form as well as its distribution nee in early Dutch is modelled after anaphoric nee, which is used as an answer to a yes/no-question. It explains why learners may initially use nee with scope over the entire clause structure. The clause in the scope of nee serves in fact as the explicit expression of a presumed yes/no-question and, hence, the meaning of nee can be paraphrased as ‘I do not want’. Since the modal operator nee has scope over the clause structure as a whole, it is referred to as ‘holistic nee’. Note, however, that nee is used mostly in utterances without a topic. Thus, unambiguous evidence of the holistic use of nee is not that frequent. Nevertheless, the holistic use of the negator is productive and is therefore considered a systematic pattern at the initial stage of acquisition. A similar use of the anaphoric negator nein has been observed for early German in (7) and (8). Here too, children use the anaphoric form to negate the utterance as a whole. (7) L1 German. Holistic ‘nein’ (1978) ‘nein’ Topic Predicate nein sauber nein hauen nein Heiko Mütze nein spielen Katze nein schaffe-ich

(no); examples from Wode (1977: 93f.) and Felix

no clean no bang no H cap no play (with) cat no do-I

(8) L1 German. Holistic ‘nein’ (no) in Benny ‘nein’ Topic Predicate, or: Topic Predicate ‘nein’ nain DA sitsa no there sit (Benny 2;2) nein mi(l)ch no milk (Benny 2;2) auto nein car no (Benny 2;2)

The same use of nee in Dutch and nein in German can be observed in adult secondlanguage learners. Examples of holistic nee in L2 Dutch and German are given in (9) and (10).





Christine Dimroth, Petra Gretsch, Peter Jordens, Clive Perdue, and Marianne Starren

(9) L2 Dutch. Holistic ‘nee’ in Mahmut and Fatima ‘nee’ Topic Predicate, or Topic Predicate ‘nee’ nee hier huis no here house (Mahmut 1.2) veel eten nee much eat no (Fatima 1.5) (10) L2 German. Holistic ‘nein’ (no) in Janka ‘nein’ Topic Predicate nein tasche eh links no bag uh to-the-left (Janka 1.3) nei platz eh *gazett* no place uh newspaper (Janka 1.3) nei eh zeitung links no uh newspaper to-the-left (Janka 1.3)

Simultaneously with holistic nee in her early child Dutch, Jasmijn (1;7–1;9) also uses the expressions unne, minne, ninne, hunne, ulle, etc. as a positive counterpart. They are productively used in precisely the same position in which nee occurs. Examples are given in (11): (11) L1 Dutch. Holistic ‘unne’ etc. in Jasmijn ‘unne’ Topic Predicate hunne Mijnie die sijfe want M that-one write (Jasmijn 1;7) unne Mijnie sijfe want M write (Jasmijn 1;7) unne Mijnie dit sijfe dit want M this-one write this-one (Jasmijn 1;7) nunne Mijnie hier sijfe want M here write (Jasmijn 1;8) unne tout poessie ballie pakke want naughty kitty ball get (Jasmijn 1;8) ulle Mijnie ook want M too (Jasmijn 1;8) ulle ik sijfe want I write (Jasmijn 1;9) ulle ukke sijfe want I write (Jasmijn 1;9)

The holistically used unne-forms seem to be modelled after the adult verb willen (want). They function as the positive alternative to nee both with respect to their distribution as to their meaning. Given that the meaning of nee and unne etc. can be described as ‘(I) do not want’ and ‘(I) want’, both devices serve to express “volition”. Other data provide additional evidence that the verb willen (want) is a positive alternative to sentence negation. The examples in (12) show both willen and its negative counterpart wil nie (from Schaerlaekens & Gillis 1987: 99). If we compare these examples with those of Jasmijn in (11) they only differ with respect to the position of willen/wilnie (want/want not) in clause-final position. (12) L1 Dutch. Holistic ‘willen’ (want), ‘wilnie’ (want not). See Schaerlaekens & Gillis (1987: 99) Topic Predicate Frans bal spelen,

‘willen’ willen F ball play, want

Finiteness in Germanic languages

papa voordoen, willen Frans pyama aandoen, wil nie

daddy show, want F pyjamas on-do, want not

At the relevant stage, Andrea uses ja ‘yes’ a few times as a modal operator to express “assertion”. Furthermore, at (2;0–2;1) Andrea uses mag-ikke ‘may-I’ systematically with the same distribution as nee. Examples are given in (13). (13) L1 Dutch. Holistic ‘mag-ikke’ (may-I) in Andrea ‘mag-ikke’ Predicate mag-ikke fomme, ja? may-I swing, yes? (Andrea 2;0) mag-ikke ijssie hebbe may-I ice-cream have? (Andrea 2;1) mag-ikke ook bank zitte? may-I too bank sit? (Andrea 2;1) mag-ikke ook gijbaan? may-I also slide? (Andrea 2;0) mag-ikke buite toe? may-I outside to? (Andrea 2;1)

Thus, Andrea seems to use both mag-ikke and nee to express “volition”. At the relevant stage, mag-ikke is used as a fixed phrase incorporating the topic element ikke. Examples of the holistic use of bitte (please) in L1 German are given in (14). (14) L1 German. Holistic ‘bitte’ (please) in Benny Topic Predicate ‘bitte’ kuche habe bib / kuche haba bi: mama cake have please mommy (Benny 2;2)

Examples of the holistic use of L2 Dutch ja ‘yes’ and niet ‘not’ to express “assertion” and ook wil ‘also want’ to express “volition” are given in (15). (15) L2 Dutch. Holistic ‘ja’ (yes), ‘niet’ (not) and ‘ook wil’ (too want) ‘ja’, etc. Topic Predicate ja die meisje brood weggooien yes the girl bread away-throw (Mahmut 2.9) ja baas politie vragen yes boss police ask (Mahmut 2.9) niet mijn familie buitenland no my family foreign country (Mahmut 2.5) ook wil altijd naar kerk too want always to church (Fatima 1.8) ook wil uh altijd goed too want – always good (Fatima 1.8)

Examples of the holistic use of bitte (please) to express “volition” in L2 German are given in (16). (16) L2 German. Holistic ‘bitte’ (please) in Janka ‘bitte’ Topic Predicate bitte komm komm wasser please come come water (Janka 1.3) bitte jacke **garderob**/ eh: please coat wardrobe(Janka 1.3)





Christine Dimroth, Petra Gretsch, Peter Jordens, Clive Perdue, and Marianne Starren

bitte bitte

tasche das deckel

zeitung komm rechts

please bag newspaper (Janka 1.3) please the lid comes to-the-right (Janka 1.6)

Summarizing, the holistic use of Dutch nee and German nein and their positive counterparts unne etc., willen, mag-ikke in Dutch and bitte in German appears to be characteristic to express volition at the initial stage of both child- and secondlanguage acquisition. Furthermore, both in L1 and L2 Dutch holistic ja and niet are used to express assertion. With respect to their form as well as their distribution some of these holistic modal operators are obviously modelled after their anaphoric use in the target language. Therefore, the meaning of holistic nee, nein can be paraphrased as ‘I do not want’, while the meaning of their positive alternatives can be described as ‘I want’. The meaning of ja and niet can be paraphrased as ‘it is true, indeed’ and ‘indeed, it is not true’. That is, an utterance containing ja (such as Mahmut’s utterances in (15) above) are more emphatic than the equivalent without ja. While the holistic use of the set of modal operators expressing volition is characteristic of the initial stage of acquisition, some of these operators may also appear in clause-internal position. This is particularly the case in later stages of acquisition when the holistic use occurs less frequently and is eventually entirely given up. The anaphoric origin of modal operators such as nee, nein, ja and bitte explains some distributional properties of holistically used expressions in early learner varieties. First, it explains why nee and nein are used as opposed to the target form of sentence negation, i.e. niet or geen in Dutch and nicht or kein in German. Second, it explains why nee and nein have scope over the clause structure as a whole. Finally, it explains why those operators of the target language that are used anaphorically such as nee, ja, and bitte, typically occur holistically in early varieties of child language.

. The Conceptual Ordering Stage: Lexical validation The term conceptual ordering refers to the fact that both the selection and the sequential ordering of constituents in learner grammar is determined by principles of information structuring. At the relevant stage of acquisition child utterances consist of three structural positions each for constituents with a particular informational function. Thus, the topic occurs in initial position. It functions as an anchoring element, i.e. it establishes external reference to the outside world or to the previous utterance. The predicate occurs in final position. It refers to a particular state of affairs which holds for the topic element. The relation between the predicate and the topic element is established by a linking element. This linking element occurs between the topic and the predicate.

Finiteness in Germanic languages

Linking devices are used to validate relations between the predicate and the topic. They constitute a closed class of lexical items which can be classified along three broad pragmatic categories: (a) the expression of positive and negative assertion, (b) the expression of assertive force in the use of scope particles, and (c) other illocutionary forces as indicated by the fixed expressions which occur as precursors for grammatical imperatives or modal verbs. As the default value of interpretation, only the positive assertion can be Left implicit. Prototypical examples are given in (17). (17) Linking devices in L1 and L2 Dutch and German L1 Dutch

Topic dit

Link nee

Predicate afdoen

this no off-do (Jasmijn 1;10) L1 German des auchnoch rausmach that too out-do (Lisa 2;0) L2 Dutch nog misschien brief komt later maybe letter comes (Mahmut 2.6) L2 German jetzt mein auch zweiundzwanzig now my brother too 22 bruder jahre years (Janka 2.1)

These examples show that elements of spatio-temporal or personal deixis occur in topic position. The topic position may be left empty if reference can be inferred from context. The closed class of lexical linking devices consists of either modal or assertive expressions (Dutch nee, ulle, ja, kanwel, kanniet, is, doettie, doemaar, niet; German ja, nein, bitte, will, soll, kann, muss) or adverbial elements (Dutch: eve, graag, wel, ook, nog; German: auch, noch, wieder). Differences in contrast are signalled via intonation. This allows learners to distinguish, for example, the claim “the predicate is true for the topic” from the claim “this specific topic is the one for which the predicate holds”. If the position of the linking device remains empty, the utterance expresses the default function of assertion. The three structural positions are ordered sequentially; syntactically, they function as adjuncts. As its name indicates, the topological structure of the Conceptual Ordering Stage mirrors the basic information-structure of the utterance. Although the Holistic Stage showed a preference towards a topic-predicate order, too, it nevertheless did not allow for a systematic expression of the relation between the element(s) talked about, the topic, and what is said about them, the predicate. At the Conceptual Ordering Stage it is now possible for the learner to differentiate between various forms of this relation between topic and predicate and, moreover, to express conditions on it by using scope particles as linking elements. So the most important step forward consists here of a lexicalised link between the topical element(s) and the predicate. Of course, the range of linking elements varies to some extent between targetlanguages and individual learners. But the main characteristics stay the same, i.e.





Christine Dimroth, Petra Gretsch, Peter Jordens, Clive Perdue, and Marianne Starren

the positional invariance of the link and the illocutionary component of interpretation which relates (links) the topic to the predicate. In the following, the different sorts of links we find during the Conceptual Ordering Stage are briefly introduced and illustrated.

.. Positive and negative assertion The Dutch examples in (18) show on the one hand the covert occurrence of the validation marker; on the other hand, its negative counterpart (niet) occurs always in between the topic and the predicate part of an utterance which justifies the assumption of a fixed position for the linking element in both cases. If the positive value of the assertion force is emphasized, i.e. if it contrasts with negation, the particle wel is used to express this contrast (see example (18) with hel; here, hel means wel). Moreover, we find overgeneralized ‘all purpose verbs’ such as doe/doet ‘do/does’ which indicate the shift towards an overtly expressed validation marker. These ‘all purpose verbs’ occur thus as precursors for analysed target-adequate finite verbs in V2 position. In all the examples, the rigidity of the positioning of the link between topic and predicate is striking. (18) Assertive validation in L1 Dutch [context]

Topic poesje die [nee Bad zitte] Ruti ik

Link 0 niet hel

Predicate vinger bijte kitty finger bite (Jasmijn 1;10) afpakke that not away-take (Jasmijn 1;11) bad zitte [no bath sit] R. indeed bath sit (Andrea 2;1) doette op zitte I do-it on-sit (Jasmijn 1;11)

In the case of German, similar phenomena can be observed. Regular positive assertion is unmarked (in the sense that no overt marker appears), its negative counterpart is net (a local variant of nicht). We also find overgeneralized ‘all purpose verbs’. In Valle’s case below in (19) this is dat ‘does’. (19) Assertive validation in L1 German Topic Link Predicate ganz hase 0 kaputt / total hare kaput / ganz hase 0 kaputt, guck total hare kaput, look (Benny 2;9) dein auto 0 kaputt fahren your car kaput drive (Benny 2;9) des net gummi this is not a rubber (Valle 1;11) mich net kitzele me not tickle (Benny 2;9)

Compare now the examples from the adult L2 learners in (20) and (21). As before, positive assertion may be left implicit, negative assertion occurs with the particles niet or nicht in linking position for Dutch and German respectively.

Finiteness in Germanic languages

(20) Assertive validation in L2 Dutch Topic Link Predicate vandaag 0 hoofdpijn today headache (Ergün 2.1) en dan 0 politie komen and then police come (Mahmut 2.9) ik niet *hapis* gaan I not prison go (Mahmut 2.9) taxi wel duur taxi indeed expensive (Mahmut 2.4) (21) Assertive validation in L2 German Topic Link Predicate Chaplin 0 gehen strasse / kommen geh spazier immer 0 alles weg in italia nix mark meine kind nix in schul

Chaplin go street / come go stroll (Janka 1.6) always everything gone (Urszula 3.2) in Italy nothing mark (Marcello 1.1) my child nothing in school (Angelina 1.1)

Distributional analysis shows that the verbs which occur in the predicate slot of the L2 learners’ utterances are not inflected but appear in a fixed, non-contrasted form. There is no indication that the verbs occur as “raised” or “made-finite” in any respect. Moreover, verbs occur optionally in assertive utterances as shown by the example Immer alles weg (always everything gone; Urszula) and In italia nix mark (in Italy nothing mark; Marcello). Finally, the topic slot is not restricted to one element only, but may host several elements which nevertheless always bear information-structural topic status. Therefore, the topic position above corresponds neither for L1 learners nor for L2 learners to a target-like topicalization position which would presuppose that one, and only one, constituent occurs in a specified pre-finite position.

.. Scope particles as links In contrast to the plain positive or negative assertion, the occurrence of scope particles as linking elements allows for a more fine-grained expression of the kind of relation that is to hold between the topic and predicate of an utterance. During the Conceptual Ordering Stage, children and adults entertain scope particles in linking function and position besides other precursors of morphological finiteness. Scope particles are a useful linking device at the Conceptual Ordering Stage. They may sometimes interfere with the acquisition of morphological finiteness at later stages of child and adult language acquisition in the following sense: as with negation, these particles have been observed to occur in complementary distribution with upcoming finite auxiliaries (see Penner, Tracy, & Weissenborn 2000 for L1 and Dimroth 2002 for L2 acquisition). Compare first the examples from L1 Dutch and German in (22) and (23).





Christine Dimroth, Petra Gretsch, Peter Jordens, Clive Perdue, and Marianne Starren

(22) Scope particles as linking elements in L1 Dutch Topic Mijnie Mijnie papa

Link ook zelf ook

Predicate heppele M. too help (Jasmijn 1;10) doen M. myself do (Jasmijn 1;10) hebbe puzzel? daddy too have puzzle? (Andrea 2;1)

(23) Scope particles as linking elements in L1 German Topic des so’n buch mama ich ich a

Link auchnoch noch auch auch auch auch

Predicate rausmach das too outmake (Lisa 2;0) such a book too (Lisa 2;0) spielplatz mommy too playground (Benny 2;2) will kaffee I too want coffee (Benny 2;2) kaffee will I too coffee want (Benny 2;2) a steigen he too in step (Valle 1;11)

The most prominent scope particle in early child language is certainly auch/ook ‘too’ (see also Nederstigt 2001 and Hulk, this volume) which appears in linking position. The additive particle expresses that in relation to a previously mentioned assertion the same predicate holds for another topical element, too. In child language, this results most often in an imperative/volitional interpretation for the respective utterances, whereas the adult data contain more additive assertions proper. This also gives semantic justification for their interpretation as linking elements. In Dutch, we find apart from ook ‘also’ the particles wel ‘indeed’ and zelf ‘self ’; in German, the particles noch ‘also’ and wieder ‘again’ occur to complete the picture during the Conceptual Ordering Stage. Of course, their individual interpretational contribution differs from the auch case in being restricted to the addition of topic times (noch) or in other dimensions depending on the respective particle. Early wieder ‘again’ appears typically in utterances with only a linking element and often in the fixed combination schon wieder ‘already again’, so topical elements and predicate are elided. This is different from the L2 learners who are more consistent in treating all the scope particles alike, i.e. as linking elements on a par with the negation operator etc. The L2 data mirror the findings of the L1 data, compare the following German and Dutch examples in (24) and (25). (24) Scope particles as linking elements in L2 Dutch Topic Link Predicate daar ook die man there also the man (Ergün 2.4) en dan nog andere politie gezien and then too other policeman seen (Mahmut 3.6)

Finiteness in Germanic languages

(25) Scope particles as linking elements in L2 German Topic Link Predicate jetzt mein bruder auch zweiundzwanzig jahre now my brother also 22 years (Janka 2.4) rote mann noch bier trinken red man another beer drink (Ivan) (der rote) wieder in de wald gegangen (the red one) again to the forest gone (Wladimir)

.. Modal expressions as links Early modal expressions occur in an uninflected, unanalysed form which is associated with a fixed illocutionary interpretation. In that way they could likewise be labelled ‘illocution marker’. They appear in the same topological position as the negation and the scope elements, in between the topic (or topical elements) and the predicate. Thus, they are linking elements by position and by their semantic/illocutionary contribution. During the Conceptual Ordering Stage these modal elements occur in a form-function correlation with interindividual systematicity as (26) and (27) show. (26) The system of modal expressions as linking elements in L1 Dutch meaning positive modal negative modal Jasmijn (1;10–1;11) ‘volition’ ulle/wil (want) nee (no-want) ‘ability’ kanwel (can-indeed) kanniet (can-not) ‘permission’ magwel (may-indeed) magniet (may-not) ‘obligation’ moet (has-to) hoefniet (has-to not) Andrea (2;0–2;1) ‘volition’ mag-ikke (may-I) nee / unnie-niet (no want) ‘ability’ kan(wel) (can-indeed) kan niet / gaat niet (can-not) ‘permission’ mag (may) magniet (maynot) ‘obligation’ hoef / moet (has-to) hoef niet / moet niet (has-to not) (27) The system of modal expressions as linking elements in L1 German meaning ‘volition’ ‘ability’ ‘obligation’ Benny (2;2/2;9) ‘volition’ ‘ability’ Valle (1;11)

positive modal will / möcht daman soll / muß, müssen will kann

negative modal – damannich – willnet kannich





Christine Dimroth, Petra Gretsch, Peter Jordens, Clive Perdue, and Marianne Starren

Whereas the Dutch modals as in (28) occur in a child-typical “proto-version” – often with the amalgamated particle wel – the German cases as in (29) appear to be closer to the appropriate target modal. (28) Modal expressions as linking elements in L1 Dutch Topic poes dit poppie Mijnie Mijnie da Jaja Peter disse

Link il nee nee kan kanniet kanniet mag moet hoeniet doettie doetie gaatie Cynthia doemaar die maa die niet

Predicate mij vinger happe afdoen ape losmake drinke melk drin zitte dop opdoen zitte meeneme alles opete hantie geve ape opmake hier doen afpakke !

kitty want my finger bite (Jasmijn 1;11) this no off-do (Jasmijn 1;10) doll no sleep (Andrea 2;0) M can loose-make (Jasmijn 1;10) M cannot drink milk it-in (Jasmijn 1;11) there cannot sit (Andrea 2;1) J may lid on-do (Andrea 2;0) P has-to sit (Jasmijn 1;11) this has-to-not with-take (Andrea 2;1) does-he everything up-eat (Jasmijn 1;11) does-he hand give (Andrea 2;1) goes-he sleep (Andrea 2;1) C do-please on-make (Jasmijn 1;11) that please here do (Andrea 2;1) that not away-take! (Jasmijn 1;11)

(29) Modal expressions as linking elements in L1 German Topic der

Link will

Predicate noch wagen haben he wants another waggon have (Valle 1;11) i will da sitz I want there sit (Benny 2;2) ich valle will möcht einen gelben I V want would-like a yellow one (Valle 1;11) nein, ich willnet sitza no, I want-not sit (Benny 2;9) nein, ich kannich soße würfel no, I cannot (throw) big dice, den damannich essen that can-one-not eat (Valle 1;11) dann ich kann bum bum dann then I can bum bum then (Benny 2;9) da daman aufmachen this can-one open (Valle 1;11) ich soll andere seite gucken I (=you) must other page look (Valle 1;11) ich soll auch hoch I (=you) must also high (Valle 1;11) des buch soll’ma / the book must one / buch anguckn book on-look (Valle 1;11) da hier müssen’se hin there here must-they away (Valle 1;11) flugzeug muß tankt werden airplane has-to refuelled be, AUTO nich car not (Valle 1;11)

Finiteness in Germanic languages

Although we could not find as many negated modals in German as in Dutch, we expect in principle no differences in this respect. This difference is probably due to the different familiarity with the interlocutor – parent-child communication in the Dutch case vs. non-parent-child communication in the German case which allows for the Dutch siblings Jasmijn and Andrea potentially easier rejection or argumentation opportunities involving negation. Depending on the target-language and the individual learner, finite content verbs enter the developmental picture earlier or later. Concerning the two targetlanguages involved, it seems that Dutch children adhere for longer to the “assertion – scope particle – proto-modals” system excluding finite verbs. This might be due to the stronger position modals have in Dutch compared to German. In L2, modal expressions likewise fill the linking position in an often similar proto-form. Examples are given in (30) and (31). (30) Modal expressions as linking elements in L2 Dutch Topic ik altijd ik vrouw

Link wil kanniet moet

Predicate zit met Nabil I always want sit with N. (Fatima 3.2) dertien uur werk I cannot 13 hour work (Mahmut 3.1) keuken woman must kitchen (Mahmut 1.7)

(31) Modal expressions as linking elements in L2 German Topic er Chaplin kind

Link will muss will

Predicate bleiben he wants stay (Janka 3.6) mit diese brief gehen C must with this letter go (Janka 3.6) telefonieren child wants phone (Urszula 3.2)

As the previous examples show, both children and adults use the topic position for expressions referring to the topic time and the topic entity. Children may initially use generalised adverbial fillers for time or place specification as German jetzt ‘now’ and da ‘there’, and even an anaphoric form indicating the move to a right-adjacent topic time dann ‘then’. Adults generally have a wider repertoire of temporal adverbs to indicate topic time. Neither type of learner yet has the linguistic means to dissociate topic time and situation time (Klein 1994, see note 5), i.e., systematic aspectual distinctions are absent from the repertoires. Within the overall restriction that topic time and situation time are equivalent, some adult learners do however manage to express, by clever use of adverbs and different types of Aktionsart, certain distinctions such as habituality and iterativity (Starren 2001), as shown in (32). (32) The expression of habituality and iterativity in L2 Dutch Topic Link Predicate altijd ik 0 wakker om 8 uur always I awake at 8 o’clock (Mahmut 1.2)





Christine Dimroth, Petra Gretsch, Peter Jordens, Clive Perdue, and Marianne Starren

drie dagen ik 0 altijd auto rijen vandaag ik

three days I always car drive (Ergün 3.2) 0 altijd weg met auto today i always gone with car (Fatima 2.6)

In the first example, one situation time is linked up to a series of topic times expressed by altijd ‘always: (habituality)’, whereas in the other two examples a complex situation time expressed by altijd ‘always’ is linked up to one topic timespan (continuity or iterativity).

. Finite Linking Stage: Grammatical validation At the Finite Linking Stage, finite verbal morphology is the typical reflex of the validation operation. Validation is obligatorily expressed through finiteness marking. It occurs via morphological adaptation of the linking element to (prominent and prototypical features of) the topic (tense, mood, person/number) and placement in a (syntactic) operator position with respect to the predicate. At this point, the story gets complicated in different respects. The morphological markings on the validation device seem to have different degrees of independence. While morphological agreement with topic time can carry temporal information alone as in Klein’s (1998) example The book was on the table, other parts of the agreement information, i.e. agreement with the external argument, seem to be less independent. With respect to topic entities, this is not easy to test, since even in prodrop languages these entities have to be explicitly referred to when reference changes. In addition to that, the validation device sometimes gets fused with parts of the predicate, namely the lexical verb, and properties of that verb (its Aktionsart for example) also then mix in with what has been described in Section 2. The transition from the Conceptual Ordering Stage to the Finite Linking Stage is characterised as a transition from lexical to grammatical validation of the utterance. For learners of Dutch and German, this transition poses three distinct acquisitional problems. Firstly, finite utterances in the target languages are marked by verb morphology, thus the “link” in non-elliptic utterances is always explicit and may contain more than one element: the morphological marking plus, optionally, one of the “linking words” from the previous stage. The learner must come to realise that this obligatory verb morphology carries both temporal/aspectual and person/number agreement information. Secondly, the finite marking falls on the leftmost verbal item, a fact which may lead to a merge between the link and the verbal part of the predicate. Thirdly, verb-second means that in declarative sentences there is only one constituent position before the finite verb.

Finiteness in Germanic languages

Table 1. The acquisition of modal and auxiliary verbs in L1 Dutch Child

age

inf. ‘doet’ / ‘gaat’ / MOD + inf. %

pp. ‘heeft’ / ‘is’ + pp. %

Jasmijn 1;10–1;11 243 107 2;0–2;2 85 134

30% 55 61% 30

2 18

4% 38%

Andrea 2;0–2;1 2;2–2;4

17% 29 50% 32

3 53

9% 62%

295 61 158 161

Given the complexity of the acquisitional task, the transition between the stages under discussion is particularly important. In both child and adult acquisition we see the task accomplished in a stepwise manner: firstly, the conceptual ordering is maintained while the learners work on the link as marked by (modal) auxiliary verbs and the copula; then the utterance-internal syntactic reorganisation takes place; finally, a choice is made between items competing for the oneconstituent topic position, which entails the development of a “landing site” for other topical items. We review the four cases of acquisition in the order in which they were presented in the previous sections. As Table 1 shows, Dutch children drastically increase their use of modal and auxiliary forms at the Finite Linking Stage. Jordens (2002) notes however that the development of auxiliaries with past participles lags behind that of modals with infinitives: whereas the latter are productive at the Conceptual Ordering Stage, the former only become productive at the Finite Linking Stage. Jordens argues that the development of the category of auxiliaries (hebben/zijn) + a past participle (pp) can be seen as the child’s way in to the grammatical tense/aspect system of the target language system. Concomitantly, gaan ‘go’ and doen ‘do’ + infinitive are developed to express aspectual values other than the perfect. These four auxiliary verbs are the first to show person/number agreement. The development of a full pronoun system accompanies this development of inflectional agreement, some aspects of which can be seen in (33). (33) The acquisition of auxiliary verbs to express aspect in L1 Dutch Topic Link ik heef die is ikke ik ik Jaja ik

Predicate afspoeld I have washed (Jasmijn 2;2) altijd opde televisie geweest that-one is always on tv been (Jasmijn 2;2) hemme deze tekend I have this drawn (Andrea 2;3) doet neusje snuite I do nose blow (Jasmijn 2;0) ga eve die glije I go just that slide (Jasmijn 2;0) doet kitkat opete J does kitkat up-eat (Andrea 2;3) doe mij fesje aan mij jamaboek I do my vest on my pyjamas do doen (Andrea 2;4) gaan’ze almaal ete, zie? go they all eat, see? (Andrea 2;3)





Christine Dimroth, Petra Gretsch, Peter Jordens, Clive Perdue, and Marianne Starren

In German too, as shown in (34) below, children at this stage show a more productive use of the modal + infinitive construction (already used at the previous stage) together with the development of the Perfekt. It is not always clear whether the value of this latter form is aspectual (perfect) or temporal (past), but it is in any case a clear sign that children now master a past, present, future directionality, i.e., more than a distinction between the here-and-now/non-here-and-now. The final example of Lisa’s in (34) shows both the use of modal wollen, and a Perfekt form with a clear aspectual value, albeit with an idiosyncratic word order. (34) The acquisition of auxiliary verbs to express aspect in L1 German Topic Link Lisa hat Alle hat da da das jetzt / Lisa

Predicate was malt en boden sauber macht is/sin lauter sachen dinne sin lauter dauda sachen dinne wollma abmachen das abmacht hat

Lisa has something painted (Lisa 2;0) Valle has the floor clean made (Valle 1;11) there is/are mere things in-there there are mere things in (=self-correction) (Valle 1;11) that want-we take off now Lisa that off taken has (Lisa 2;00)

As soon as the auxiliaries heb/heeft/hat, gaa(t) and doe(t) have come to be used to express aspectual distinctions, the relation between the elements in second position and the predicate has been reinterpreted, too. When children have established a grammatical relation between auxiliary verbs, on the one hand, and the predicate on the other, they have, in fact, discovered the relation between the structural position of a head and its complement. Whereas the illocutionary elements of the conceptual ordering stage are adjuncts, auxiliary verbs are part of a functional category system. As is the case in the target language, they function as the head of a head-complement structure. Those lexical-semantic and grammatical-syntactic properties which are relevant to this process of restructuring are given in (35). (35) From ‘Illocutionary force’ to the function of a ‘Head’ conceptual ordering stage: proto MOD grammatical status: lexical category meaning: illocutionary force syntactic function: adjunct

finite linking stage: AUX functional category grammatical aspect head

At the same time, the acquisition of morphological person/number agreement with an external argument is evidence of the acquisition of a specifier-head relation between the NP and the auxiliary. Having acquired both the head-complement and the specifier-head relation with auxiliary verbs the children have learned the gram-

Finiteness in Germanic languages

matical properties of the functional category system which determines the projection of the inflectional phrase in Dutch and German. Thus, the acquisition of both the head-complement and specifier-head relationship can be traced to the acquisition of the auxiliary verbs hebben/haben and zijn/sein in the context of past participle forms as a means to express perfect (and/or past, for German) meaning. With the adult learners too, it is only from the grammaticalized finiteness phase on that formal agreement with contextual information appears in the linking slot, in the form of tense and person marking, first on auxiliary, and ultimately on lexical verbs. Examples are given in (36) and (37). (36) Finiteness marking on non-thematic verbs in L2 Dutch Topic Link Predicate hij is weggelopen he is away run (Ergün 2.9) dan is die meisje ook hier then is the girl also here sleep (Ergün 2.9) slapen ik moet nog opereerd I must still operated (Mohamed 3.4) (37) Finiteness marking on non-thematic verbs in L2 German dann wir

habe freundschaft then we have friendship make/made mache (Tino 1.4) in diese strasse war auch Chaplin in this street was also Chaplin (Janka 3.6) der Charlie hat auch gemacht the C has also made (Cevdet 3.6) du musst nicht so mache you must not so do (Tino 2.7) sie habe nix merke she have nothing notice(d) (Casco 3.2)

As shown in (36) and (37), the negator is placed to the right of the auxiliary verbs and to the left of lexical verbs (see also Becker, to appear). Particles and adverbs, such as ook, auch ‘also’ and nog ‘still’, are progressively integrated into the utterance structure, starting to appear behind auxiliary verbs. Their domain of application is no longer necessarily adjacent. In In diese strasse war auch Chaplin ‘in this street was also C.’ auch is left-adjacent to its domain of application Chaplin, whereas in Der Charlie hat auch gemacht ‘the C has also made’ its domain of application is the non-adjacent Der Charlie. As the ultimate step to the Finite Linking Stage, children’s grammars appear to develop a specifier position within the VP constituent, which was not available at the Conceptual Ordering Stage. As a result of this, the initial position becomes available as a possible landing site for constituents such as the internal argument or an adverbial element, while at the same time the external argument occurs in the specifier position of VP. This syntactic phenomenon of topicalization now accounts for the anchoring function of items in the initial position. However, the placing of different items which are informationally topical in different positions in accordance with the verb-second rule of the target languages takes time to master.





Christine Dimroth, Petra Gretsch, Peter Jordens, Clive Perdue, and Marianne Starren

Mastering finiteness and verb-second is not a one-step development, as the following examples show. Underspecification of the topic occurs in (38). Valle (1;11) was with his mother (M) at the station in the morning and saw a man missing a train: (38) Underspecification of the topic in Valle (1;11) V: der bahnhof is pät kommen the station is late arrived M: der Bahnhof? the station? V: nich der bahnhof sondern der mann not the station but the man

‘The man’ and ‘at the station’ are topical elements, but Valle does not manage to integrate both into the utterance. Overspecification of the topic occurs in (39). Valle (1;11) is building a house; the adverb (j)etzt ‘now’ is used twice, once before, once behind finite is. (39) Overspecification of the topic in Valle (1;11) V: etz hier is jetzt niks dinne now here is now nothing in-it

For adult learners, as with the children, mastering finiteness and verb-second is not a one-step development. The examples in (40) show that two constituents may precede the finite verb: (40) Finite verbs in V3 position in L2 Dutch and German Topic Linking Predicate dan die polizei hat ihn festgehaltet then the police has him fast-held (Cevdet 2.6) toen een meisje heeft brood gepakt then a girl has bread grabbed (Ergün 3.9)

To summarize, with the mastery of the Finite Linking Stage for children and for adults, if they get that far, positive assertion is expressed by finiteness marking on a verbal element. Morphologically, finiteness also marks the link to the topic time TT, as well as person/number agreement to the grammatical subject. For negative and qualified assertion, the negative and additive particles come to be used in conjunction with verbal morphology. Different verb forms indicate tense (i.e., whether the topic time TT for which the assertion is made occurs before, at, or after the time of utterance TU), and different verbal combinations indicate grammatical aspect (i.e., whether the topic time TT concerns the whole, or just part, of the temporal structure of the predicate: Tsit).

. Summary of the stages of development As a result of an analysis based on an information structural approach, striking similarities were observed in the acquisition of Dutch and German by children

Finiteness in Germanic languages

learning their mother tongue and adults learning a second language. A common property of these different learning situations is the expression of a topic and a state of affairs, such that the state of affairs is claimed to hold for the topic. This relation between the topic and the state of affairs is established through what we have termed a validation or linking device. It is this validation device which is expressed differently at consecutive stages of language acquisition, but where each stage shows striking cross-linguistic and cross-subject similarities. These different ways to express validation allowed us to discriminate between a Holistic Stage, a Conceptual Ordering Stage and a Finite Linking Stage. At the Holistic Stage validation is mainly achieved pragmatically, i.e., validation typically depends on variables inherent to the discourse. At the Conceptual Ordering Stage validation is achieved lexically by the use of elements expressing illocutionary force. At the Finite Linking Stage validation occurs morpho-syntactically by the use of elements with a grammatical function. Characteristic of the Holistic Stage of acquisition is the use of modal operators in clause-initial or clause-final position. These modal operators are used holistically, i.e. they have scope over the topic-predicate structure as a whole. The holistic use of clausal operators occurs most prominently with nee ‘no’ in early child Dutch and nein ‘no’ in early child German. Both forms have a negative modal meaning as in ‘I don’t want’. As positive alternatives Dutch children use elements based on the target verb form wil(len) ‘want’ or mag-ikke ‘may I’, while in German children use will ‘want’ and bitte ‘please’. These modal expressions are used meaning ‘I want’. With similar distribution expressions such ook wil ‘too want’ and bitte ‘please’ can also be found in adult L2 learners of Dutch and German. At the Conceptual Ordering Stage basic utterance structure consists of a sequence of three structural positions. Constituents in these positions are related by adjunction, while their ordering depends on principles of information structuring. Each position can be filled by more than one constituent, which means that target-language syntactic constraints have yet to be acquired. Constituents in initial position function as topic elements. A topic element can be an NP, often a proper name, or a deictic adverb. Topics have anchoring function, i.e. they establish the embedding of an utterance in context. Modal elements occur in second position. They belong to a closed-class category of modal expressions that are used with illocutionary force. They allow both children and adults to express volition, ability, possibility, obligation and assertion. Since assertion is the default pragmatic function, the position of the element used to express assertion may be left empty. At the conceptual ordering stage, modal phrases are lexical linking devices with no syntactic function. Therefore, scope particles such as eve ‘just’, wel ‘indeed’, ook ‘too’ and zelf ‘self ’ may also occur in second constituent position. Finally, the constituents in end position are termed predicates. A predicate can be a VP or a VP-like expression, referring to a particular state of affairs. The relation between the two





Christine Dimroth, Petra Gretsch, Peter Jordens, Clive Perdue, and Marianne Starren

constituents is established by the linking element. It serves as a validation device, i.e. it is used to express that a particular state of affairs as referred to by the predicate holds for a particular topic. Validation may vary with respect to the kind of relation that is claimed to hold. Validation of the relation between the state of affairs and the topic element is grammaticalized at the Finite Linking Stage. Elements of the target functional category of auxiliaries are used to express both aspect and illocutionary force. Due to the fact that the finite verb has come to be used as a grammatical linking device functioning as the head of a head-complement structure, scope particles and zero marking cannot be used any more as a means of validation. The same holds for nee, nein ‘no’ and other target adverbial elements. Furthermore, the acquisition of morphological person/number agreement with an external argument is evidence of the acquisition of a specifier-head relation between the NP and the auxiliary. Having acquired both the head-complement and the specifier-head relation with auxiliary verbs children and adults have learned the grammatical properties of the functional category system which determines the projection of the inflectional phrase in Dutch and German.

. Conclusion Finiteness is a functional property of Germanic languages. It is carried by auxiliaries and lexical verbs occurring in second constituent position. In order to explain processes of the acquisition of finiteness it seems necessary to distinguish between the semantic concept of finiteness and its formal representation (Klein 1998). The semantic concept of finiteness entails two properties of information structuring: anchoring and linking. Anchoring is the pragmatic operation which establishes the identification of what is talked about and the embedding of the actual utterance in a discourse world. Linking is the operation which validates the relation between a particular state of affairs described in the predicate part of an utterance and a topic element. While both functions are fused in one finite verb form in the target languages investigated, they are clearly separated in learner varieties of both children and adult learners. Evidence from the analysis of first and second language data shows that the acquisition of these information structural properties of finiteness occurs in three consecutive stages of linguistic development: the Holistic Stage, the Conceptual Ordering Stage and the Finite Linking Stage. At each of these stages of language acquisition the relation of linking or validation is expressed linguistically differently. At the Holistic Stage it is achieved by pragmatic means, at the Conceptual Ordering Stage by lexical means and at the Finite Linking Stage by morpho-syntactic means.

Finiteness in Germanic languages

The devices of validation turn out to be remarkably similar in children learning their mother tongue and in adults learning a second language acquisition. At the Holistic Stage, frequently no validation device is expressed at all. We often find some topic and predicate in a juxtaposition relation and even predicates occurring alone. Characteristic however, is the use of a few modal operators with scope over the clause structure as a whole. These modal operators are typically elements such as nee ‘no’ or ulle ‘want’ in Dutch and nein ‘no’ or bitte ‘please’ in German. They are used with anaphoric function in the target language model. At the Conceptual Ordering Stage child and adult utterances consist of three structural positions each for constituents with a particular informational function. The topic occurs in initial position. It establishes external reference to the outside world or to the previous utterance. The predicate occurs in final position. It refers to a particular state of affairs which holds for the topic element. The relation between the predicate and the topic element is established by a closed class of linking elements. These linking elements occur between the topic and the predicate. Linking devices used to validate relations between the predicate and the topic, constitute (a) elements expressing positive or negative assertion such as doe(t) ‘does’, wel ‘indeed’ or niet ‘not’ in Dutch and dat ‘does’ or net ‘not’ in German (b) scope particles such as ook ‘also’, nog ‘too’ and zelf ‘self ’ in Dutch and auch ‘also’, noch ‘too’, wieder ‘again’ in German and, finally, (c) modal phrases expressing volition, ability, possibility, and obligation, such as mag-ikke ‘may-I’, nee ‘no’ , kanwel ‘canindeed’, kanniet ‘cannot’, magwel ‘may-indeed’, magniet ‘may-not’, moet ‘has-to’, hoefniet ‘has-to-not’ in Dutch and will ‘want’, nein ‘no’, kann ‘can’, kannich ‘cannot’, danman ‘may-one’, dananich ‘may-one-not’, soll or muss ‘has-to’ in German. At the Finite Linking Stage validation of the relation between the state of affairs and the topic element is grammaticalized. Acquisition of the auxiliaries heb, heeft, ben, is ‘have, has, am, is’ etc. in Dutch and hab(e), hat, bin, is(t) ‘have, has, am, is’ etc. in German leads to a reanalysis of modal expressions with illocutionary force as members of the category AUX. It establishes a syntactic category which functions as the head of a head-complement relation. Furthermore, the morphology of the elements of AUX may also represent the anchoring function of spatio-temporal reference and personal deixis. As such it establishes a specifier-head relation, too. Both the head-complement and the specifier-head relation are grammatical properties of the functional category system of the INFL-phrase in Dutch and German. Finally, the standard assumption of the differences in the expression of finiteness between child and adult acquirers is that the former quickly and effortlessly develop the necessary verbal morphology and the latter do not. Here, we have tried to give a more nuanced picture, showing that there are more similarities in the process than is usually assumed.





Christine Dimroth, Petra Gretsch, Peter Jordens, Clive Perdue, and Marianne Starren

Notes * We are most grateful to Katrin Lindner for many helpful comments on an earlier version of this paper. A preliminary version of this paper was given as a poster presentation at the EuroConference on Information Structure, Linguistic Structure and the Dynamics of Acquisition. The Structure of Learner Language. Sant Feliu de Guixols, Spain, 7–12 October 2000. . See also Lasser’s (1997) distinction between S(emantic)-finiteness and M(orphological)finiteness. . This idea is standard in enunciative approaches. Culioli (1995 [1983]: 106), for example, states (English translation of Michel Liddle): “In an assertion, we can see that this [predicative relation]. . . will be located with respect to . . . a system of spatio-temporal co-ordinates, as well as the subject of the utterance. This will enable me to say that such and such a relation is validated for a specific moment. . . ”. . Note that the term predicate is not used in its syntactic/categoric sense. . “prise en charge” in enunciative terms, cf. Culioli (1995). . Klein (1994) contrasts this topic time – TT: the time for which a claim is made – with the time of situation – Tsit, corresponding to the temporal properties of the predicate. Grammatical aspect (see below, §2.3) is then defined as the relation between these two time spans. . To some extent this is also true for the topic place. We don’t want to claim that reference to a topic place is in the same sense an obligatory part of the topic information as reference to the topic time, but thetic sentences especially often seem to require place reference: “there is a flower” is probably not an acceptable English sentence. Still very often, the topic information has both meaning components. Unfortunately, we cannot go further into this question here.

References Becker, A. (to appear). The semantic knowledge base for the acquisition of negation and the acquisition of finiteness. In H. Hendriks (Ed.), The Structure of Learner Varieties. Mouton: de Gruyter. Culioli, A. (1995). Cognition and Representation in Linguistic Theory. Texts Selected, Edited and Introduced by Michel Liddle. Amsterdam: Benjamins. Dimroth, C. (2002). Topics, assertions and additive words. Linguistics, 40, 891–923. Felix, S. (1978). Linguistische Untersuchungen zum natürlichen Zweitspracherwerb. München: Fink. Hendriks, H. (2000). The acquisition of topic marking in L1 Chinese and L1 and L2 French. Studies in Second Language Acquisition, 22, 369–397. Hulk, A. (this volume). Merging Scope particles. Word order variation and the acquisition of aussi and ook in a bilingual context. Jordens, P. (2002). Finiteness in early child Dutch. Linguistics, (40), 687–765. Klein, W. (1994). Time in Language. London: Routledge.

Finiteness in Germanic languages

Klein, W. (1998). Assertion and finiteness. In N. Dittmar & Z. Penner (Eds.), Issues in the Theory of Language Acquisition (pp. 225–245). Bern: Lang. Klein, W. & Perdue, C. (1997). The Basic Variety. Or: Couldn’t natural language be much simpler?. Second Language Research, 13, 301–347. Lasser, I. (1997). Finiteness in Adult and Child German. Doctoral Dissertation, City University of New York. Available through Max-Planck-Institute of Psycholinguistics in Nijmegen. Meisel, J. (1991). Principles of universal grammar and strategies of language use: On some similarities and differences between first and second language acquisition. In L. Eubank (Ed.), Point-Counterpoint: Universal Grammar in the Second Language (pp. 231–276). Amsterdam: Benjamins. Meisel, J. (1997). The acquisition of the syntax of negation in French and German: Contrasting first and second language acquisition. Second Language Research, 13, 227– 263. Nederstigt, U. (2001). Prosody: A clue for the interpretation of the additive focus particles auch and noch. Linguistische Berichte, 188, 415–440. Penner, Z., Tracy, R., & Weissenborn, J. (2000). Where scrambling begins: Triggering object scrambling at the early stage in German and Bernese Swiss German. In S. M. Powers & C. Hamann (Eds.), The Aquisition of scrambling and cliticization (pp. 127–164). Dordrecht: Kluwer Academic Publishers. Perdue, C., Benazzo, S., & Giuliano, P. (2002). When finiteness gets marked: The relation between morpho-syntactic development and use of scopal items in adult language acquisition. Linguistics, 40, 849–890. Schaerlaekens, A. & Gillis, G. (1987). De taalverwerving van het kind. Een hernieuwde oriëntatie in het Nederlandstalig onderzoek. Groningen: Wolters-Noordhoff. Schlyter, S. (this volume). Development of verb morphology and finiteness in children and adults acquiring French. Skiba, R. (1993). Funktionale Analyse des Spracherwerbs einer polnischen Deutschlernerin. In A. Katny (Ed.), Beiträge zur Sprachwissenschaft, Psycho- und Soziolinguistik. Probleme des Deutschen als Mutter-, Fremd- und Zweitsprache (pp. 201–225). Rzeszów: WSP. Starren, M. (2001). The Second Time. The Acquisition of Temporality in Dutch and French as a Second Language. Utrecht: LOT. Wode, H. (1977). Four early stages in the development of L1-negation. Journal of Child Language, 4, 87–102.



On the similarities of L1 and L2 acquisition How German children anchor utterances in time Petra Gretsch

.

Introduction

When faced with the problem to anchor utterances in time, (Germanic) languages provide us with three strategies to master this task: (i) leave it implicit to be specified via context or default interpretation, (ii) specify a morphological tense/aspect marking, or (iii) choose a temporal adverbial/an explicit temporal expression. As for the first account, there is no linguistic burden put on the speaker which makes it a perfect starter strategy for L1/L2. But the expressive power of this account is obviously very limited in that the speaker, i.e. the learner, exerts no control over the flow of temporal information. Concerning the morphological (ii) and the adverbial account (iii), an opposing sequencing of development has been proposed for L1 and L2: Whereas in L1 an early morphology (before 3–4yrs.) is supplemented with late adverbials, L2 shows the reverse pattern with early adverbials which may later get complemented with morphological markers if the L2 learner passes a certain developmental point which has been labeled the stage of ‘Basic Variety’ (see Klein & Perdue 1997; for a comparison of temporal reference in L1 and L2 see also Hendriks 1999; Starren 2001). In contrast to Hendriks (1999) who analyzed the anchoring of utterances in connected discourse, this paper investigates episode-initial time anchors in freely occurring, self-initiated narratives from three longitudinal German child corpora1 and compares their emergence with the onset of morphological tense/aspect markers in addition to an overall comparison of the acquisitional sequence in L1 and L2. This paper aims at changing this coarse picture of opposing developments in arguing that the morphological strategy and the adverbial strategy emerge simultaneously in L1. Which kind of strategy is the first in the L1-data to be manifested depends on the individual learner’s path to acquisiton w.r.t. the integration of verbal morphology. Moreover, the opportunities given for self-initiated topic time set-



Petra Gretsch

ting in the conversations with adult interlocutors play a role as well and render the comparison of interindividual corpora, in particular child corpora problematic. It has also been shown that the characteristics of the source language and the targetlanguage exert to some extent an influence on the development and the usage of temporal markers as shown a.o. by Berretta (1990).

. Mechanisms of temporal anchoring . Target language properties The main question of this paper asks how learners anchor events in time. This is best tackled in relation to the notion of topic time as introduced by Klein (1984), which lies at the core of the linguistic description of temporal parameters pursued here. Topic time reflects the necessity to anchor utterances within a time (and possibly also place) matrix to render utterances interpretable in the first place. As an example for a topic time specification consider the following question-answer pair: (1) A: Why couldn’t Mary deliver her talk yesterday? B: Because she was ill. The use of the past tense marker in B’s answer implies that Mary was ill at the crucial “topical” time interval pointed to by yesterday in combination with the time span covered by her expected talk as expressed in the antecedent question. But nothing has been said about Mary’s health condition today – she could just as well still be ill, or recovered from her illness. Thus we need to distinguish between the situation time (TSit) in B’s utterance ‘Mary’s being ill’ and the topic time (TT) of the utterance, i.e. the time span about which a claim is made which is in the above case ‘the time slot of her expected talk yesterday’. In more recent papers of Klein (1998, to appear) it is claimed that it is the finiteness element which is the grammatical expression of assertion and which serves as a link between TT and the sentence basis which is asserted “over it”. In general, I assume a tripartite topic specification of topic entity (the standard topical element, i.e. the person/entity talked about), topic time and topic place. None of these topical elements requires syntactically to be overtly expressed but they require then to be contextually inferrable (see also Gretsch & Dimroth 1999). Within the Germanic languages the morphological expression of tense is obligatory intertwined with this expression of finiteness (at least for all tensed, i.e. finite sentences). The occurrence of tense itself restricts thereby the array of potential temporal interpretations depending on the text type. But as can be seen from the example above, tense is not the only factor: A complex interplay of factors enables the calculation and interpretation of the temporal time spans involved which occur within a sen-

On the similarities of L1 and L2 acquisition

tence (including the most prominent one which is relevant for the assertive claim, i.e. the topic time). This can include information on tense and aspect as well as adverbial or nominal temporal specifications in combination with rules of temporal ordering in and across discourse. In the following I concentrate on the close relationship between adverbial TT markers and morphological tense expression as it shows up in restrictions on their combination. For the overall temporal interpretation, the relation between the implicitly or explicitly given topic time (adverbial, NP, PP) and the morphologically given tense value constitute a crucial factor. The interplay of this sort of “tense-agreement” is exemplified in the following examples where a TT-expression marking ‘posteriority’ (put in square brackets) is combined with a present tense morphology: (2) a. *[Gestern] Yesterday b. *[Gestern] Yesterday

ißt eats fällt falls

Maria einen Apfel Maria an apple ein Apfel vom Tisch. an apple from-the table

(3) a.

[Im Jahre 1579] ißt Maria Stuart einen Apfel In-the year 1579 eats Maria Stuart an apple ‘In the year 1579, Maria Stuart is eating an apple.’ b. ?[Im Jahre 1579] fällt ein Apfel vom Tisch In-the year 1579 falls an apple from-the table ‘In the year 1579, an apple is falling from the table.’ c. [Vor vielen Jahren] wird in Altdorf ein Apfel gespalten, Ago many years becomes in Altdorf an apple torn, der für die Schweizer. . . that for the Swiss. . . ‘Many years ago, an apple is torn in Altdorf which turned out to be a major event for the Swiss.’

The violation of the agreement relation as exemplified under (2) blocks at first any coherent interpretation. This is different in the well-known cases of historical or narrative present tense (see (3)) where the temporal inconsistency between TT-marker and tense morphology can be interpretationally resolved and pragmatically exploited: In the above case, by conveying immediacy/directness in the rendering of past events. The occurrence of historical names as ‘Maria Stuart’ or places which are famous for a historic event (‘Altdorf ’ and its ‘Apfelschuß’ from Wilhelm Tell) can combine with potential other TT-markers to allow for a complex TTspecification. This TT-set comprises in the case of (3c) e.g. (TopicTime: vor vielen Jahren, TopicPlace: Altdorf/Schweiz). If an interpretation as topic and with it a TTinterpretation of the respective historical indicator is impossible, the interpretation





Petra Gretsch

of the present tense morphology as historical present tense is untenable or odd. This is e.g the case if the historic TT-indicator occurs in a clefting construction. Here, the interpretation of Maria Stuart as a focussed constituent allows only for a correction reading, compare (4a, b) with (4c): (4) a.

?Es

ist Maria Stuart, die (im Jahre 1579) einen Apfel ißt It is Maria Stuart who (in-the year 1579) an apple eats ‘It is Maria Stuart, who is eating an apple (in the year 1579).’ b. ?Es war Maria Stuart, die (im Jahre 1579) einen Apfel ißt It was Maria Stuart who (in-the year 1579) an apple eats ‘It was Maria Stuart, who is eating an apple (in the year 1579).’ c. Es ist/war M.S., die (im Jahre 1579) einen Apfel aß It is/was M.S. who (in-the year 1579) an apple ate ‘It is/was Maria Stuart, who was eating an apple (in the year 1579).’

The precise conditions which allow for inconsistent tense-agreement are particularly intricate to define since we have to deal here with a conglomerate of pragmatic factors as text type and underlying quaestio of a discourse (see (4a)), information structural rules which interfere with topological ordering and their interplay with a (not obligatory overtly expressed) TT-marker (compare (5a, b), see also (6a, b, c)). vor Gericht: ?[Gestern] ißt Maria einen Apfel, zwei Stunden at-the court: Yesterday eats Maria an apple, two hours später ist sie tot. later is she dead ‘Yesterday, Maria is eating an apple, two hours later she is dead.’ b. vor Gericht: *Maria ißt gestern einen Apfel, zwei Stunden at-the court: Maria eats yesterday an apple, two hours später ist sie tot. later is she dead

(5) a.

(6) a.

[Während Jascha (*gestern) Kirschen pflückt], aß Maria While Jascha yesterday cherries plucks, ate Maria (gestern) einen Apfel. yesterday an apple ‘While Jascha is plucking cherries, Maria was eating an apple.’ b. [Während Jascha (gestern) Kirschen pflückte], ißt Maria While Jascha yesterday cherries plucked, eats Maria (*gestern) einen Apfel. yesterday an apple ‘While Jascha was plucking cherries, Maria is eating an apple.’

On the similarities of L1 and L2 acquisition

c. *Maria ißt einen Apfel, [während Jascha Kirschen pflückte]. Maria eats an apple while Jascha cherries plucked ‘Maria is eating an apple, while Jascha was plucking cherries.’ Note that the interpretation of (6a) and (6b) is the same in localizing the event at some point in the past. Due to the required simultaneity of the event of plucking and the eating-event as encoded by während the possible readings are narrowed down to past tense readings of the main TT, i.e. TT < TU and [TT = TSit1] = TSit2. But the interpretational steps in how this reading comes about differ in the two cases with noticeable effects on the possibilities of additional TT-elements according to tense-agreement. If we want to add the TT-marker gestern to the two sentences, (6a) allows gestern to occur only in the matrix clause, whereas (6b) restricts its occurrence to the subordinate clause. If the während-clause follows the main clause as in (6c), it looses its TT-selecting force to the effect that an inconsistent tense-agreement can not be pragmatically resolved. Moreover, the two issues of grammatical aspect and the temporal characteristics of the verb itself (Aktionsart) complicate further the picture. Since I focus on TT-markers and tense in this paper I omit the discussion on these interfering factors here. I assume following Klein (1984) that it is tense that relates the TT to the time of utterance whereas aspect specifies the relation between TT and TSit which therefore does not reveal any direct bearing on the time span of validity for an assertion. Since this paper focuses on the temporal markings of learner varieties in relation to the time of utterance, only topic time markers and tense markings play a role to this end. To sum up, this shows that the calculation of the relations among the temporal variables in a sentence reveals a complex system in the target language involving different levels of linguistic analysis. Moreover, the requirement of tense-agreement as a phenomenon with structural and interpretational effects leads to the expectation that the development of adverbial markers and that of morphological markers of temporal relations are related in learner varieties, too.

. Learner language properties – general options In consequence, this renders the overall task of temporal interpretation and the issue of tense-agreement in particular a complex and difficult one for learners. But a learner can do with much less target grammatical background to convey more or less gross temporal specifications. These can nevertheless signal the temporal span of validity of an assertion, i.e. the TT. There is no agreement relation necessary to produce and comprehend TT-markers at the starting point of acquisition if there is at least one element to anchor the temporal interpretation onto. The three options available to a learner are: 1. TT-specification by contextual clues only (implicit TT)



 Petra Gretsch

2. TT-specification by morphological means and2 3. TT-specification by adverbials or NPs The first option, the contextual one, requires the least lexical, morphological and/or grammatical knowledge for production, but a correspondingly greater effort for interpretation. But this is only possible on the cost that this simple strategy restricts the moments of time which can be talked about to the fixed ‘here&now’ or to otherwise contextually controllable moments which have been already established by the interlocutor. A typical example from child language is the following, illustrating this latter contextual option in a non-here&now context: (7) Lisa02 (2;0.17) L: da there (pointing to a mosquito-bite on her hand) Ad: Was hastn da gemacht? ‘What have you done there?’ L: mücke domt (=kommt) mosquito comes Ad: Hat die dich gestochen? ‘Has it bitten you?’ L: ja. ein mückentich Yes. A mosquito-bite Although no explicit posteriority marker was used in the first turn in (7), the adult interlocutor is able to infer from the mosquito-bite on Lisa’s hand that she is in the post-state of being bitten by a mosquito which allows the adult to ask ‘What have you done there?’. This makes the shifted TT to a past event explicit and allows Lisa in her reply to pick up on this shifted TT with – in principle – virtually any inflectional form of a verb. As the use of the proto “present tense marking” on the verb kommt in Lisa’s answer shows, do verbal markings or other TT-specifying devices not matter for signaling the intended temporal interpretation as long as the TT-anchoring is already positively negotiated between the interlocutors. The morphological (ii) and the adverbial account (iii) contrast sharply in the amount of grammatical knowledge involved: The morphological strategy presupposes the category verb to be equipped with some paradigmatic distinct marking. For it to be applicable the learner has to build into his or her grammatical system knowledge about the mechanism of morphological tense/aspect marking in general and their associated meaning for a morphological TT-specification. Moreover, tense-marking implies the issue of finiteness in Germanic languages. Together with the interplay of topological and scope-induced positioning rules this constitutes a highly complex learning task if we broaden the morphological aspects to

On the similarities of L1 and L2 acquisition 

the morpho-syntactic level. It is even more complex for learners with German as target-language, where the tense-marked verb has to appear in second position (if it is the main verb). This means that the set of topical elements (TT, TopicPlace and TopicEntity), requires a decision as to which of its elements appears in the initial, syntactically defined “topicalisation” position, which is grammatically restricted to one element only (and might also host a focal constitutent instead of a topical one in contrast to its label). Thus, the 1:1-correspondence of conceptual organization and formal organization breaks down as soon as “tensing plus Vfin2” is acquired for learners of German (see also the description of the conceptual ordering stage vs. the finite linking stage of L1 and L2 learners in Dimroth et al., this volume): Table 1. Structural Linking needed for the morphological strategy

Hence, acquiring a finitely-marked verb has an array of complex structural ramifications which have to be mastered in order to conform to target grammar. See the following examples, taken again from child language, which exhibit an early morphological paradigm at the age of 2;1 for the indication of different TTs: here&now with present tense (8a) vs. future/volitional force expressed with the infinitival marking (8b) and past events/perfectivity with tensed auxiliary plus participle in target-adequate places (8c): (8) Valle06 (2;1.7) a. da kommt der Schaufelbagger hin there comesfin the dredger prt ‘There we place the dredger.’ b. ein bild noch draufmalen a picture prt on-paintinf ‘(The adult shall) paint a picture on it.’

 Petra Gretsch

c.

valle hat den auspuff gebaut valle hasfin the exhaust-pipe built ‘Valle has built the exhaust pipe.’

Compare this with the much simpler “adverbial strategy”. All the learner has to acquire, besides the lexical information, is how to attach an adverbial or any other topic-time-indicating phrase to his or her grammatical structure. This is done by adjunction as can be shown by “satelliting structures” where the topical element(s) can either occur to the left or right of the sentence kernel. No verbal morphology needs to be involved, no topological problems occur: Table 2. Structural Linking needed for the adverbial strategy

Assuming a topic-focus conceptualisation based on the focus-last principle (for the default case of having the topical information fronted), the linguistic structure simply mirrors the conceptual structure: First present the topical anchor(s), then the sentence part which is predicated about it. The more elaborate the lexical material in expressing TT-specifications, the greater is the expressive power of the adverbial option. Take for example the following utterances from adult second language which exploit the above sketched topological frame with its possibilities of placing temporal adverbials either in topic or in predicate position to convey specific aspectual distinctions (see Starren 2001). The TT part is again marked by square brackets: (9) Turkish-Dutch: Ergün [drie dagen ik] altijd auto rijen three days I always car drive ‘For three days, I was always driving the car.’ But despite the inclusion of aspectuality, the adverbial option has nevertheless its limitations: TT and TSit are still undissociable. The inclusion of temporal adverbials of contrast like still, already etc. into a learner’s lexicon presupposes a finite, i.e. a morphologically equipped sentence structure as has been shown by Perdue et al. (2000). So, the range of tense/aspect distinctions of the adverbial option excludes complex temporal relationships with a dissociation between TT and TSit by its lack of inflectional means. And as the case of the adverbials of contrast shows, the set of temporal adverbials available to a learner is also dependent on the devel-

On the similarities of L1 and L2 acquisition 

opment within the realm of morpho-syntax and the development of inflection in particular. Depending on the learner variety and the individual learner strategy, we observe also a mixing of strategies concerning the morphological and the adverbial option. The contextual-only option which occurs in all learner and also in nonlearner (i.e. target-speaker) varieties serves anyway as a default option with different interpretations of TT depending on the text type and overall quaestio of the text for an unvariable TT (as e.g. in descriptions) or a monotonic TT-shift (as e.g. in narratives). In contrast, the morphological option and the adverbial option are assumed to figure in a distributed fashion corresponding to the respective learner variety. The general view on acquisition has it that L1 learners adhere to the morphological option whereas L2 learners “choose” the adverbial option due to their late integration of inflectional markings. If we consider the amount of grammatical knowledge involved, it is the more surprising that children should start with the morphological option before they learn about temporal adverbials. It is noteworthy that the expressive power of the two strategies obviously differs drastically in quality. Depending on the learner’s progress with the integration of the system of inflectional markings, the morphological option embodies in principle the whole range of tense/aspect distinctions but excludes by its limitations in its expressive power any specific (adverbial!) temporal specifications. In contrast, the adverbial option allows for a very detailled TT-specification as long as no other complex relationships between TU, TT and TSit are involved (see the case of temporal adverbs of contrast) and renders the additional tense/aspect marking thus somehow superfluous. In the following I inspect the alledged distribution of L1 learners following the morphological route and L2 learners following the adverbial route more thoroughly starting with the case of L2 first.

. Characteristics of L2 The analysis of developmental progression of untutored L2 learners is based on data from the ESF-Corpus (for an analysis of ESF-Data concerning the issue of temporal interpretation see also Klein & Perdue 1997; Starren & van Hout 1996; Perdue et al. 2000 and Starren 2001).3 For the comparison between L1 and L2 to be as close as possible the text-types which have been taken into account comprise personal narrations on the one hand and story-retellings (Charlie Chaplin: ‘Modern Times’) on the other. Positional temporal adverbials are among the earliest topic time markers learned in L2 as illustrated in (10) and (11). Up to the stage of the socalled Basic

 Petra Gretsch

Variety, a prominent stage of acquisition where a lot of learners fossilize, inflection markers do not play any role in the specification of topic time:4 (10) Turkish-Dutch: Ergün [vandaag] hoofdpijn. ik ga niet naar fabriek today headache. I go not to factory ‘Since I have a headache today I am not going to the factory.’ (11) Moroccan Arabic-French: Zahra [mon mari les vacances], Maroc l’accident my husband the holidays Marroco the accident ‘During the holidays, my husband had an accident in Marocco.’ During the stage of the Basic Variety, the interpretation of adverbials is controlled by their placement. Due to the overall influence of the focus-last principle, temporal adverbials are either interpreted as TT-markers if they appear at the left periphery of the utterance or as TSit-markers if they appear within the predicate part (the non-topical part) of the utterance at the right periphery: Time adverbials may occur in utterance initial position, most clearly in narratives. An utterance in the main structure (the ‘foreground’) of a narrative answers a ‘quaestio’ such as What happened at time tx ? Thus, the topic of a foreground utterance contains a time span tx , and the focus is the event that happened at that time. Therefore, a time adverbial specifying the time span of the ‘quaestio’ occurs naturally with pattern PH3 [these patterns are phrasal constraints observed in the Basic Variety; PH3 is defined as giving the order VNP or Cop-NP. PG], giving Adv-V-NP. A background clause, by contrast, may answer an implicit question such as When did this happen? In this case, it is the specification of the time span which is in focus and, hence, an adverbial which specifies this time span comes in final position. (Klein & Perdue 1997: 318)

A related point can be found in Stutterheim and Klein (1989) where differences in the positioning of temporal adverbials are related to their main structure vs. side structure status in the discourse. Although the expression of topic time involves only a subset of temporal adverbials at the stage of the Basic Variety (mostly positional and deictic temporal adverbials), the L2 learner has already achieved a system of temporal relationships between three time parameters: the time of utterance (contextually always given), the topic time as expressed in the “topicalised” elements like vandaag, and les vacances, and the time of the situation (TSit) as expressed in the rightward predicate part of the utterance. The restriction is that if TT and TSit occur both in an utterance, the TT-specification serves as a temporal indicator for TSit in setting TT = TSit. This allows only a broad simultaneity interpretation for TT and TSit, since grammaticalized tense or aspect expressions are still lacking.

On the similarities of L1 and L2 acquisition 

Nevertheless, the more refined system of temporal reference which is gained by the combination of temporal adverbials within an utterance makes an aspectual system possible within the limits of simultaneity interpretation, see Starren and van Hout (1996) and Starren (2001).The multiple use of adverbials thereby partly allows the learner to compensate for the (still) missing tense/aspect morphology: (12) Zahra [moi quatre mois] toujours fatigue I four months always tired ‘For four months I was always tired.’ (13) Mahmut [altijd ik] wakker om acht uur always I awake at eight o‘clock ‘I always wake up at eight o‘clock.’ (14) Mahmut [gisteren-jaar] altijd ongeluk gedaan yesterday-year always accident done ‘Last year I always had accidents.’ The minimal pair of positional contrast of altijd in example (13) and (14) makes the difference between topic time and situation time interpretation particularly clear: Whereas (13) has to be paraphrased as ‘It is always the case (= TT) that I wake up at eight o’clock (= TSit)’ and thus conveys a habitual interpretation, the utterance under (14) conveys an interpretation like ‘As for the last year (= TT), I always had accidents (= TSit)’, i.e. an iterativity interpretation. The verb does not play any prominent role for the temporal anchoring of utterances here. The creativity of adult learners in expressing concepts without having the means of the grammatical target is demonstrated nicely by the use of gisteren-jaar ‘yesterday-year’ in (14) where the learner paraphrased the expression last year in a very individual manner. It is only in highly advanced stages that learners use an inflectional based target-like tense/aspect system as illustrated by the target-adequate (15) which also exhibits a clefting construction proper: (15) Gloria c’est pas la fille qui a volé le pain It-is not the girl who has stolen the bread. This stage of a grammaticalized linking element (see also Dimroth et al., this volume) may still involve multiple topic elements in front of the finite marked verb (16) or the lack of an obligatory subject-agreement as in (17) – here due to the missing of the grammatical subject itself. This shows that the expression of finiteness in learner language is apriori dissociated from the V2-rule and from the

 Petra Gretsch

(limited) obligatoriness of having an overt subject and thus an overt relation of subject-verb agreement: (16) rg-05 [dann er] hat im restaurant gesessen then he has-been in-the restaurant sitting ‘Then he has been sitting in the restaurant.’ (17) cg-02 [dann] is wieder eingeschlafen then is again sleeping ‘Then (s/he) has been sleeping again.’ In contrast, the rule of tense-agreement is respected as soon as the inflectional means are acquired. As said before, we encounter substantial changes during the developmental progression from the BV to a stage implying grammaticalized finiteness/tense. This is expected if we assume the relation of tense-agreement to have noticable effects in L2, too. In sum, the developmental path for the expression of temporal markers in L2 includes the following steps: 1. Spatio/temporally interpretable nominals (en vacances etc.) and positional temporal adverbials like yesterday, last week as earliest TT-markers. 2. The topological space of sentences is exploited for the differentiation of TT vs. TSit (depending on the information-structural status of the temporal variable involved). 3. The combination of two temporal adverbials allows for the expression of (still simultaneity-bound) aspectual differences as e.g. iterativity and habituality. 4. The combination of temporal adverbials and grammatical marking appears with the integration of a grammaticalized tense/aspect system and allows for a new system of temporal combinatorics and agreement patterns. 5. Contrastive temporal adverbials appear which involve a relation between three independent temporal parameters like still (where TSit is valid for two different topic times); their structural integration depends crucially on the development of a grammaticalized expression of finiteness. The steps from (1) to (3) characterize the Basic Variety stage with (3) already indicating an orientation towards a grammaticalization of complex temporal relations. The two last steps (4 and 5) entail the usage of inflectional means and a grammaticalized expression of finiteness in particular. But this might still entail among other deviances from the target structure a non-target-adequate placement of the finite verb (see again example (16) above).

On the similarities of L1 and L2 acquisition 

Even if we grant different target languages to have a different influence on the speed with which morphological information gets included into the L2-learners system, the overall coarse progression-line seems nevertheless to be a quite stable one: First temporal adverbials and NPs carrying temporal information occur, then inflectional morphology (if it occurs at all). But it is also crucial to note that the development of the range of temporal markings as expressed within the adverbial option is not independent from the development of inflectional tense marking / finiteness marking. On the basis of the grammatical tense-agreement as discussed in the section on the target language, this relationship is expected.

. Characteristics of L1 The data for the L1 part of this study consist of three longitudinal corpora of monolingual German children covering an age range from 1;11 to 4 years (see Note 1) in addition to data from an elicitation study with 33 children (as available in the CHILDES database). The search for topic time markers in child language is complicated by the fact that most of the spontaneous interactions in longitudinal studies are adult-child communication where the adult specifies the topic time via asking all sorts of questions and therefore bans self-initiated topic time markers on the side of the child. As shown by Ninio and Snow (1996), it generally holds that children do hardly discuss matters of the non-present at all: In their study, the token frequency of these utterances ranges from 1.7% at the age of 12 months, to 6.4% at the age of 24 up to 7.6% at the age of 32 months (loc. cit. p. 113).5 Interestingly, this seems to mirror the input data from the mother: [. . . ] the discussion of the nonpresent shows an almost perfect match of mother and child distributions thoughout the period studied. (loc. cit. p. 119)

Deviances from the discussion of the here&now either comprise of volitional acts (which still bear a strong link to the here&now in having an actual wish or need at the time of utterance) or they comprise of rare instances of discussions of past events. Methodologically, I tackled this problem by focussing on child-initiated topic time changes, and the topic time specification involved in setting the stage for a personal narrative in particular which gives us a “past” interpretation. All child-initiated topic-time shifts were categorised according to the grammatical means involved which then allowed in a further step for a developmental generalisation across the three longitudinal studies. What counts as a topic time shift was determined co- and contextually. Apart from the ubiquitious contextual-only topic time specification as already exemplified under (7), children make an early use of morphological differentiations to convey different TTs and with it differences in world-referencing. This is illustrated in the following episode:

 Petra Gretsch

(18) Valle03 (1;11.25). Mother S2 and child V are narrating their visit to the city of Reutlingen earlier this morning to a second adult interlocutor S1 whilst V is playing with farm toys. S2: Weißt noch was da gmacht ham? (. . . ) ‘Do you remember what we have done there?’ V: eisdaufen ice-cream-buyinf [past reference] S2: Ein Eis gekauft, ja! ‘An ice-cream (we) bought, yes!’ S1: (. . . ) Was warn des fürn Eis? ‘What kind of ice-cream was it?’ V: ell nomal eis essen ?? [Valle/wants] again ice-cream eatinf [future] S2: Nomal Eis essen? Du, wenn wir wieder in die Stadt kommen. ‘[You want to] again ice-cream eat? Well, if we get back to the city.’ (V throws over the basket with the farm toys) V: alle macht Valle makesfin [present] V: sütet alles alles um poursfin everything everything out [present] Typically, it is the adult interlocutor – here the mother – who sets a non-present TT in her first turn. The use of the infinitive points to a temporal setting distant from the here&now whereas the use of the finite verbs macht and schüttet show a clear relation to the present. Although the first reply of Valle, eiskaufen, could also be interpreted as an antecedent ellipsis, this interpretation is definitely excluded in the second infinitival reply which does not qualify as an answer to the question of S1.

.. Range of TT-expressions in L1 As soon as adverbial TT-markers occur, some children encounter mapping problems which are typical for child language only. The problems are conceptually or structurally motivated or both. These problems occur if two or three members of the whole set of topical elements (TopicTime, TopicEntity and TopicPlace) change simultaneously as is often the case when a new topic is introduced. Then, more than one topical element competes for the topicalization position in “being the leftperipheral topic” (see (19)). In this example we find a mismatch of topic mapping due to a simultaneous change in the spatio/temporal topic element (Bahnhof ‘station’) and the topic entity (Mann ‘man’) at the beginning of the self-initiated narration of a past event.

On the similarities of L1 and L2 acquisition 

(19) Valle03 (1;11.25). Valle was with his mother at the station this morning; he saw a man just missing his train. V: der bahnof is pät dommen the station is late arrived S2: Der Bahnhof? ‘the station?’ V: nich der bahnhof sondern der mann not the station but the man S2: Nicht der Bahnhof, sondern der Mann ist zu spät gekommen. ‘not the station but the man has arrived to late’ In addition, children have to cope with the overall conceptualization of topical and non-topical information. Problems in this area lead to the phenomenon of being over- or underexplicit in the expression of topical elements and TT in particular. This is exemplified in (20) where the overmarking of explicit reference to the default ‘here&now’-setting with jetzt and hier leads also to a target-inadequate positioning of elements in the sentence, i.e. a violation of the V2-rule: (20) Valle03 (1;11.25) S2: Soll ma de Dach machn? Des hat Krach gemacht so is des. ‘Shall we do the roof? That has made noise so is it.’ V: etz hier is jetzt niks dinne now here is now nothin in-it If we compare the cases of (19) and (20) which stem from the same file, the mapping problem under (19) could also be interpreted as a processing problem. Whereas the overmarking of topical elements as in (20) appears in all three corpora, cases of explicit mapping mismatches (19) are exceptional.

.. Variety of TT-marking in L1 The above cases of German child language exhibit all an inflectional marking on the verb and thus early attempts on grammatical tense. The question is now whether morphological tense marking precedes adverbial tense marking in general in child language. In two of the three child corpora this is indeed the case, but the corpus of the third child, Benny, shows the opposite progression: His developmental path starts with adjoined TT-markers and only later integrates an inflectional system for the marking of tense. Consider the following example where a past time reference has been induced by the sole use of the topic-setting NP hause oma (‘granny’s home’):

 Petra Gretsch

(21) Benny02 (2;02.15). B had soup at his granny’s house the day before; his mother is serving a soup right now. B: hause oma suppe au home granny soup also ‘When we were at granny’s home yesterday, we had soup, too.’ M: Wenn mer bei der Oma zuhause sind, dann gibt es immer Suppe. ‘If we are at granny’s house then we always get soup.’ In contrast to the corpora of Valle and Lisa, Benny is a comparably slow learner with hardly any verbal elements at all in his proportion of the lexicon of the first files. The complete set of (potential) verb types in all of their occurring verbal forms at this stage consists of the elements (sis ‘shoot’, heia ‘go-to-bed’, schnarcht ‘snores’, will ‘want’, [nei]sitze ‘[in]-sit’, trink ‘drink’, isse ‘is + unclear contracted pronominal form’, macht ‘makes/made’, and habe ‘have’).6 These nine verb types contain also three cases of imitations from the input. The use of these verb forms point to no correlation between form and pragmatic function in this file. Obviously, this child has not yet distinguished between the finite and the infinitival marking at all and the different verbal forms are ignored. This is in line with his blurred pronounciation of verbendings in general and the occurrence of the still contracted precursor form isse for the copula. None of the verbal forms occurs in an utterance context distant from the here&now with the exception of will pointing to a volitional speech act. In addition, this early employment of will ‘want’ shows that the distinction between here&now and distant from here&now can either be expressed via the insertion of will as an early marker for volitional speech acts and thus for a distant TT-time (Benny) or via an inflectional distinction on the verb (see the development of Valle and Lisa). Bennys development is thus close to the development of children learning English, where an early use of want and wanna-contractions in particular have been observed. (22) Benny02. (2;02.15). M. is making coffee. B: ich au will fee. ich auch fee will. I also want coffee I also coffee want To sum up Benny’s development, Benny uses the adverbial option first before he starts to exploit the inflectional paradigm for temporal marking. Moreover, he shows a considerable freedom in the placement of verbal elements, V2 is not acquired yet (see example (22)). Considering another corpus, the child Lisa, I observed an interesting scopemarking phenomenon in her use of temporal adverbials. Her early development of TT expressions is very similar to Valle’s path with early morphology and later adverbials. But as soon as her utterance length allows for the occurrence of multiple adverbials, she tends to mark the scope of the TT-marker overtly by bracket-

On the similarities of L1 and L2 acquisition

ing the event under discussion with a “scope-opening” and a “scope-closing” TTadverbial. This phenomenon even carries over into the discourse domain where her scope-bracketing spans over several turns; compare the examples (23) and (24): (23) Lisa11 (2;4.23) L: jetzt räum ich das aus jetzt now take I this out now (24) Lisa11 (2;4.23) L: manch manchmal als ich mich hier an.. dem großen Bett some sometimes when I me here at the big bed und den en dem kleinen and the en the little und dann and dem Teppich dann häng ich so runter dann so\ and then and the carpet then hang I so down then so\ Ad: Toll kannst du turnen! ‘You are very good in exercising.’ L: und dann lieg ich so auf dem Teppich\ und dann turn ich and then lie I so on the carpet\ and then exercise I manchmal so rein\ sometimes so in\ (. . . ) und dann klep ich – kletter ich so drüber manchmal (. . . ) and then clim I – climb I so over sometimes dann\ then\ L: ‘Sometimes I am exercising the following way: I hang down from the big bed and from the small bed, then back to the carpet.’ A: ‘You are very good in exercising!’ L: ‘And then I lie on the carpet in that way. And then I exercise that way sometimes.’ Since jetzt ‘now’ in (23) does not occur in a TT-shifting discourse situation, it is the more surprising that the default TT here&now is twice marked as a special case of overmarking. The occurrence of the second jetzt after the structurally sentence-closing verb particle aus indicates the satelliting, structure-peripheral nature of these scope-bracketing devices. In the case of (24), the habitual aspect of the action-sequence is expressed via the temporal adverbial manchmal ‘sometimes’ and it includes the use of several (und) dann ‘(and) then’ as an overt marking for TTshifts in a personal narrative to convey a whole series of exercising actions within the habitual frame. When she adds another part to this sequence of actions in her second turn, she closes the frame of habituality again with manchmal to mark the





Petra Gretsch

status of the described (and simultaneously performed) action as belonging to the same sequence. To complement these generalisations from spontaneous production I also examined the occurrence of temporal adverbials in an elicitation task. Children were asked to retell a story as depicted in a picture book.7 Since the story contained several breaks or overlappings in topic time it was particularly tailored to elicit TT markers to achieve a coherent retelling. The following table gives an overview of frequencies and prototypical adverbials from three different age groups: Table 3. Frequencies and prototypical temporal adverbs from story-retelling age

# subj.

type:token ratio

selection of prototypical adverbials

3y 5y 9y

12 chil. 11 chil. 10 chil.

16:101 = 0.2% 16:182 = 8.8% 27:272 = 9.9%

wieder, jetzt, dann, noch, schon, immer, gleich as above plus bald, zuerst, am nächsten Morgen as above plus eines Nachts, in der Zwischenzeit, morgens, einmal, abends

As the table above indicates, contrastive anaphoric adverbials like noch ‘still/yet’ and schon ‘already’ are among the first temporal adverbials learned and constitute the most widely used in child language. Judged by their target-semantics, contrastive anaphoric adverbials are much more complex than positional adverbials since the former can involve a relationship between three temporal parameters (time of utterance, time of the situation and topic time) whereas the latter involve only two of them. What the table does not show is that wieder ‘again’ and noch ‘yet/still’occur even earlier in child speech around 2;0. In contrast to that, it is only from four years onwards that children come to use positional adverbs like the next morning etc. So the question is why children start to tackle the more complex temporal adverbials first? An answer to this is given in the next section where I compare L1 with L2 development.

.. General options in L1 Generalising across all three longitudinal corpora of spontaneous child language we find the following two major paths for the development of temporal markings with some permeability and commutability between successive steps due to the interindividual variation:8 Path A (two corpora): 1. Basic partitioning of the spatio/temporal space into here&now as expressed by finite verbs and distant from here&now as expressed by infinitival marked verbs (covering thus past and future temporal reference). 2. Basic adverbial pointer to the here&now with jetzt ‘now’, and hier ‘here’.

On the similarities of L1 and L2 acquisition

3. (a) Basic adverbial pointer to the spatio/temporal space distant from here&now appear creating the opposition pairs hier-da ‘here-there’, and jetzt-dann ‘nowthen’. (b) The introduction of left-adjacent nominal elements for topic time anchoring. (c) In Lisa’s case first overt scope-marking occurs (bracketing phenomena). 4. Further partitioning of the tense-space covered by inflectional morphology with the introduction of participles on the one hand and a variety of modals on the other hand. 5. Introduction of syntagmatic present perfect constructions. 6. Introduction of other temporal adverbials besides jetzt and dann (esp. wieder ‘again’, noch ‘still/yet’). 7. Introduction of positional temporal adverbs like gestern ‘yesterday’ or am Abend ‘on that evening’ etc. Path B (one corpus): 1. Basic adverbial pointer to the here&now with jetzt ‘now’, and hier ‘here’ occur. 2. (a) Basic adverbial pointer to the spatio/temporal space distant from here&now appear plus (b) the introduction of left-adjacent nominal elements for topic time anchoring (as in the oma hause/ suppe au\ example) plus (c) the volitional marker will ‘want’. 3. Partitioning of the spatio/temporal space into here&now expressed by finite verbs and distant from here&now expressed by infinitival marked verbs (covering past and future temporal reference). 4. Further partitioning of the tense-space covered by inflectional morphology with the introduction of participles on the one hand and a variety of modals on the other hand. 5. Introduction of syntagmatic present perfect constructions. 6. Introduction of other temporal adverbials besides jetzt and dann as above. What these two sequences of acquisition reveal is that the linguistically expressable partitioning of the conceptual time line develops from a coarse two-way distinction between here&now and distant from here&now to a more and more fine-grained partitioning within the realm of morphological and adverbial differentiation. For those children who choose the “L2-way of temporal development” as Benny, the TT marking allows from early on for a pointer to a specific temporal moment compared to those children who choose first the morphological/grammaticalized option of TT-expression. But in both L1-paths, positional adverbials hardly play any role at the beginning of temporal development.





Petra Gretsch

The general possibility to follow Path B in child language allows for two conclusions: (a) there exists no apriori split in a morphological option reserved for L1-learners and an adverbial option for L2 learners which might be imprinted into the two varieties in their acquisitional process and (b) the intrinsic interplay of adverbial and morphological expressions of time might serve as a mutual “trigger” in both acquisition processes. As a result the adverbial and the morphological option of TT-specification do not constitute separate routes of development but form an acquisitional continuum within which children and to some extent also adults can find their individual paths towards a (more or less, in the case of L2) target-adequate realization of TT.

. Comparison of L1 and L2 To sum up the empirical description of L1 and L2 development, two striking questions await an answer: A. Why do some children prefer the complex tense/inflection system of topic time marking over a structurally much more simple mechanism which adjoins adverbials or topic time-indicating nominals to the left of an utterance? B. Why do children learn the more complex contrastive anaphoric adverbials first before they get to the semantically simpler positional adverbials? The question under (A) has been already partly answered: The adverbial and the morphological option do not form two independent routes to acquisition but conspire in development as well as in the phenomenon of tense-agreement in the target language. On the other hand, the morphological starting point embodies also an actual advantage over the adverbial one – at least for children: The gross partitioning of here&now vs. distant from here&now concerning the conceptualization of the temporal space in children is directly reflected in the two-way split of early verbal morphology (finite vs. infinitival). The more fine-grained the partitionings get the more inflectional forms are integrated into the child’s system. Since temporal adverbials do not occur as salient opposition pairs in the same way as the verbal morphology does, they are less qualified to represent the child’s basic partitioning of the temporal space. Note that the early inflectional split into finite vs. infinitival marked verbs does not entail the acquisition of the complex tense(/aspect) system. But the usage of the inflectional split for the expression of opposite topic times marks the first step to this end. The crucial question to ask is then – again – why is this particular inflectional distinction for children so salient, wheras adult L2 learners do not seem to care about it (for a longer or shorter time)?

On the similarities of L1 and L2 acquisition

Regarding the question under (B), it should be noted that this paper focuses on the topic time anchoring of individual utterances which amounts to the topic time anchoring of a single event at least in the case of L1. There is no comparison between the topic times of other events involved as long as verbalisations of events are not temporally related to each other in coherent discourse. In contrast to this, L2 learners start their acquisition process already equipped with knowledge about discourse coherence and the cognitive prerequisite of the interpretation of temporal relations. It is precisely the relation of grammatically introduced temporal parameters across utterances which is extremely difficult for the child and which might partly account for the differences encountered in L1 and L2 acquisition. The child’s specific deficit in establishing anaphoric relationships across utterances manifests itself among other things in the restriction of child utterances to the description of one event or one state per utterance (see also the examples above) in addition to the lack of overt (temporal) connectors as before, after, while etc.9 So, the TT-shifter und dann serves more as a passepartout TT-marker in narrations as an actual productive TT-link between and across utterances. Since it is the task of temporal adverbials to relate events in the verbal domain they constitute a particularly hard problem to acquire. More precisely, it is the meaning of the temporal adverbials and not the structural operation of their integration via adjunction which poses a challenge for L1 learners as can be seen in the case of Benny. He allows for early TT-adjunction with a temporal specification expressed by an NP in topic-typical fronted position. Morphological tense-marking on the other hand, does not require any comparison between events as long as tense and aspect are conflated in the early system of morphological time reference which distinguishes only coarsly between future, present and past reference.

. Conclusion The phenomenon of tense-agreement in the target language of German points to a close relation between leftperipheral expressions of topic time and grammatical tense marking. Therefore, it is expected that this relationship is also reflected in learner languages. The comparison between L1 and L2 learners has shown that the acquisition of verbal temporal markers follows a gobbled pattern which does not fit into the general view of L1 learners following “the morphological route” to temporal marking and L2 learners following “the adverbial route”. Instead, we find with respect to child language that both options are available and used by children. So, they mark only tendencies of strategic choices and not two distinct paths in acquisition as reflected in the intermingled successive steps to target-adequate tense-agreement. Concerning the adult learners, they exhibit a sharp bias towards





Petra Gretsch

the adverbial option – within the ESF corpus no learner has been found to start with the morphological option – but tense-agreement shows also effects for them. So, the development of the set of temporal adverbials of contrast is coupled onto the development of inflectional morphology which reflects differences in semantic complexity of temporal adverbials in general. If we take these results together, it can be concluded that the assumed bifurcation between the processes of L1 and L2 acquisition is to some extent a surface phenomenon which is not based on apriori differences in the possible gateways to grammar. It is rather the conceptualization discrepancy in addition to the multiplicity of fossilization stages in L2 which sets L2 acquisition apart from L1 instead of radical different paths of access to grammar.

Notes . The data are taken from the Tübinger Corpus of child language within the DFG-project “Erwerb der komplexen Syntax” under the header of Dr. R. Tracy. It comprises in total 5 monolingual and 5 bilingual corpora which have been collected for 4–5 years on a bi-weekly basis. Most of the sessions have also been video-taped. . See Gretsch (submitted) for an account of the interpretation of early finiteness markers in child language which is based on corresponding cell-divisions for the development of conceptual and grammatical means to express temporal anchoring. . For a detailed description of the data collection process and the data base of the European Science Foundation corpus on second language acquisition see Perdue (1993). . For a detailled description of this stage and its relevance in L2 development see Klein and Perdue (1997). . The number of subjects ranges from 7 to 24 per age-group. The sessions were done on a bi-monthly basis leading to 11 age groups within an age range of 12 mths. to 32 mths. The numbers are based on a pragmatic analysis of speech use of a 30 min. interactive session. The category ‘Discussions of the Non-Present’ were one category out of 7. . The age of the child is 2;2.15, the duration of the session was approx. 30 min. and comprises 292 child utterances in total. . These are retellings of the frog-story in the CHILDES corpus. I am very grateful to B. Schmiedtová for her help with the analysis. . For an account on variation in child language, the ‘building-block’ model of acquisition, see d’Avis and Gretsch (1994) and Gretsch (2000). . E.g. an influential study on the description of events in time by Clark (1974) starts with data collection at the age of 3;6 onwards.

On the similarities of L1 and L2 acquisition

References Beretta, M. (1990). Il ruolo dell’infinitivo nel sistema verbale di apprendenti di italiano come L2. In G. Bernini & A. Giacalone Ramat (Eds.), La temporalitá nell’accquisizione di lingue seconde (pp. 51–80). Milano: Franco Angeli. Clark, E. (1974). How young children describe events in time. In G. B. Flores d’Arcais & W. J. M. Levelt (Eds.), Advances in Psycholinguistics (pp. 275–284). Amsterdam: NorthHolland. d’Avis, F.-J. & Gretsch, P. (1994). Variations on Variation: On the Acquisition of Complementizers in German. In R. Tracy & E. Lattey (Eds.), How tolerant is Universal Grammar? (pp. 59–109). Tübingen: Niemeyer. Dimroth, C., Gretsch, P., Jordens, P., Perdue, C., & Starren M. (this volume). Finiteness in Germanic languages. A stage-model for first and second language development. Gretsch, P. (2000). Fokale Ellipsen in Erwachsenen- und Kindersprache. Tübingen: Niemeyer. Gretsch, P. (2000). What does finiteness mean to children? A crosslinguistic account on root infinitives. [submitted to Linguistics] Gretsch, P. & Dimroth, C. (1999). Topictime, finiteness and informationstructure: Ansätze zum Realismus. Manuscript, Max-Planck-Institut für Psycholinguistic, Nijmegen. Hendriks, H. (1999). The acquisition of temporal reference in first and second language acquisition: what children already know and adults still have to learn and vice versa. Psychology of Language and Communication, 3(1), 41–59. Klein, W. (1984). Time in Language. London/New York: Routledge. Klein, W. (1998). Assertion and Finiteness. In N. Dittmar & Z. Penner (Eds.), Issues in the Theory of Language Acquisition. Essays in Honor of Jürgen Weissenborn. Bern: Lang. Klein, W. (to appear). On finiteness. In V. van Geenhoven (Ed.), Semantics meets acquisition. Dordrecht. Klein, W. & Perdue, C. (1997). The Basic Variety. Or: Couldn’t natural languages be much simpler? Second Language Research, 13(4), 301–347. Ninio, A. & Snow, C. E. (1996). Pragmatic development. Boulder, CO: Westview Press. Perdue, C. (Ed.). (1993). Adult language acquisition: Cross-linguistic perspectives (Vol. I: Field Methods). Cambridge: Cambridge University Press. Perdue, C., Benazzo, S., & Guiliano, P. (2000). When finiteness gets marked: the relationship between morpho-syntactic development and use of scopal items in adult language acquisition. [submitted to Linguistics] Starren, M. & van Hout, R. (1996). Temporality in learner discourse: What temporal adverbials can and what they cannot express. Zeitschrift für Literaturwissenschaft und Linguistik, 26(104), 35–50. Starren, M. (2001). The Second Time. The Acquisition of Temporality in Dutch and French as a Second Language. Utrecht: LOT. von Stutterheim, C. & Klein, W. (1989). Referential Movement in Descriptive and Narrative Discourse. In R. Dietrich & C. F. Grauman (Eds.), Language Processing in Social Context (pp. 39–76). Amsterdam: North-Holland.



Negation and relational predicates in French and English as second languages Patrizia Giuliano

.

Introduction

This paper investigates the acquisition of negative particles together with the development of verbal morphology, in two Spanish speakers learning French and two Italian speakers learning English. All informants come from the ESF Project (cf. Perdue 1993). On the basis of the results of Giuliano (in press), we will present, on the one hand, the stages leading adult second language learners to mark the relational information on the verb which contributes to the notion of finiteness (temporal and aspectual anchoring; subject/verb agreement). On the other hand, we will describe the development of the expression of negative functions. Our final purpose is to show the interaction between the emergence of relational, or non-thematic, predicates: the copula, have/avoir as possession predicates, avoir/être and be/have as temporal auxiliaries, and modals) and the mastering of postverbal and postauxiliary negation. Thus, we shall concentrate exclusively on sentence negation,1 namely on the French structure (ne +) Cop/Vlexical/Mod/Aux + pas and the more diversified English constructions Cop/Mod/Aux + not and do/does/did/will. . . + not (symbols and terminology are given in detail at the end of the paper).

. Previous research The debate about the acquisition of negation by adult second language learners is concerned with the potential relation between the morpho-syntax of negation and the morpho-syntax of the verb. It draws much of its interest from the fact that in L1 acquisition the marking of finiteness on the verb represents a reliable test to evaluate the mastering of negative syntax (cf., for instance, Clahsen 1988a, b; Meisel

 Patrizia Giuliano

1997a, b). When children master verb morphology, they also place the sentence negator correctly with respect to the VP. In a larger perspective, the debate on the acquisition of negation is focused on what is available and what has to be acquired in first and second language acquisition, that is to say on the initial state of the learner (for a discussion of the initial state, cf. Eubank & Schwartz 1996). For adult learners of a second language, in particular, there is disagreement about the availability, at the beginning of the acquisitional process, of relational notions (or “functional categories”, in the generative perspective) such as tense, aspect and subject-verb agreement. The disagreement does not only involve different approaches against each other, but also does so for the tenants of the same approach. For generativists, the relevant issue is to find out whether adult learners have access to functional categories, via L1 (cf. Schwartz’ & Sprouse’s 1996, Full Access/Full Transfer Hypothesis), or whether they have to develop these categories again just like children acquiring the mother tongue, even though not exactly in the same manner (cf., for instance, Lardière 1998; and Schlyter 1999). In the functional perspective – to which our study subscribes – the question is phrased differently. First of all, functionalists distinguish between the notion of inflexion and that of finiteness. This distinction is crucial for the question of the initial state in adults. Inflexion refers to the development of morpho-syntactic markings, which, at least in Indo-European languages, are affixed on the verb or anyway englobed in the verb group; on the contrary, finiteness does not refer to a languagespecific grammar but rather to the cognitive availability of the notion of the validity of an assertion for a time span and/or a topic-subject and/or a spatial landmark. Now, for adult second language learners, the cognitive availability of the notion of finiteness is not to be questioned; nevertheless, the typological distance/proximity between the source language (SL) and the target language (TL) could involve the “re-acquisition” of such a notion or of some of its features. This would be the case when the conceptualisation of tense and/or aspect and/or agreement were different in the languages in question. However, if these languages present remarkable similarities, or even a complete overlap, in the way they conceptualise tense, aspect and agreement, then the acquisitional problem is just a matter of acquiring the right formal means for the expression of these notions. On the basis of work published in the last decade (Bernini 1996; Stoffel & Véronique 1996; Becker & Dietrich 1997; Meisel 1997a, b; Schlyter 1998a, b; Parodi 1998; Bardel 2000), there is no agreement yet as to the principles guiding adult acquisition of negation and verb morphology in an L2. One school of thought sees the nature of the adult acquisition of negation lying in strategies of input segmentation that are independent of the finite or non-finite character of the verb, and in pragmatic strategies (the negator would be placed after the verb to be close to the item it affects; cf. Meisel 1997a, b, for German L2; and Bardel 2000, for Italian

Negation and relational predicates

L2). This perspective is also the one that stresses differences between first language acquisition and second language acquisition. Another school of thought tries to show that, on the contrary, the acquisition of negation in an L2 is indeed tied up to the acquisition of verb morphology and syntax (cf. Becker & Dietrich 1997; Parodi 1998; Giuliano, in press). Under this second perspective, the acquisitional process of an L2, even though regarded as essentially different from the acquisition of a first language, can go through stages that come close to the stages observed in children. Whatever the perspective adopted, a common observation is that in adults there is variability in the marking of finiteness on verbs; this variability is absent in children.

. Theoretical framework and objectives The framework we have adopted for our analysis is the functional approach of Klein and Perdue (1992) and the notions they propose to describe the acquisitional path of adult second language learners. Within this framework, the authors distinguish the prebasic variety (or nominal utterance organisation), the basic variety (or non-finite utterance organisation) and the postbasic variety (or finite utterance organisation). In our analysis we shall concentrate on the transition from the basic to the postbasic variety, since this transition shows us the way adult second language learners go when acquiring means to mark relational information (tense, aspect, subject-verb agreement) on verbs. According to Klein and Perdue “finite utterance organisation” refers to the marking of finiteness on verbs. We shall therefore not use this definition to refer to utterances containing markings for tense or aspect other than the morphological means supported by verbs.2 Some more remarks are necessary. The extremely long and tortuous path through the postbasic stage (PBV) sometimes forces us to refer to an “initial” or “final” postbasic variety without, however, adopting two different acronyms. The initial postbasic variety is a transitional stage where learners begin to mark finiteness on the verb. This variety shows some functional oppositions but the emergence of functional morphology at this stage means that it is “functional” in the learner’s interlanguage, not necessarily in the target language (cf. §5 for a further discussion of this point). The final postbasic variety shows a larger range of functional oppositions in the domain of verb morphology. These oppositions become more and more similar to the ones native speakers mark by the same morphological means.3 As to our objectives, we shall try to demonstrate that non-lexical verbs, and in particular be/être and have/avoir (both as simple predicates and auxiliaries) play a prototypical role in marking finiteness on verbs. We shall also point out that the



 Patrizia Giuliano

sentence negator always follows these verbs. Finally, we shall try to explain this state of affairs on the basis of the non-thematic nature of be/être and have/avoir, which we shall generally refer to as relational predicates. The argumentation about non-thematic verbs that we propose has already been formulated, in the generative approach, by Parodi (2000). Parodi reanalyses ZISA4 data and, in agreement with the theoretical framework she adopts, she suggests a syntactic explanation of what she observes with respect to verb morphology: with Italian and Spanish learners of German, finiteness emerges by subject-verb agreement marking on functional predicates, namely by a feature whose character is strictly syntactic. Indeed, Parodi explicitly limits her definition of finitenessmarking: “this present study aims to show that verb placement is related to finiteness, understood as subject-verb agreement” (2000: 356). However, in contrast to Parodi, we think that the relational nature of non-thematic verbs plays a much more complex role in learner varieties, since they are prototypically used to mark tense and aspect as well.

. The informants Our Spanish informants, Bernarda and Palmira, come from Chile; the Italian speakers, Santo and Andrea, come from Southern and Northern Italy, respectively. Bernarda and Palmira were forced to move to France for political reasons; Santo and Andrea moved to England in order to find a better job. All informants had relatively few contacts with the L2; all of them had a little schooling. Bernarda and Palmira were interviewed from the very beginning of their stay in France (Bernarda after 1 month, Palmira after 2 months); Andrea and Santo had been living for a longer period in the target country (Andrea 7 months, Santo 9 months) when the ESF data collection started. The data collection was organised into three cycles, each one lasting for approximately eight months. At the onset of the inquiry, Bernarda’s and Palmira’s competence in French was very low (the prebasic variety stage of acquisition cf. Klein & Perdue 1992). Andrea and Santo were already in a basic stage; yet prebasic features were still present in their production. All the learners progressed to a postbasic variety, except Santo, whose development stopped at the basic variety. The state of affairs just reported can only partly be explained through the personal history of the informants: all of them, except Santo, had stable commitments in the target country. But Santo had more frequent contacts with the L2!

Negation and relational predicates

. From Spanish to French . Negation in Spanish Spanish belongs to the subgroup of Romance Languages that Bernini and Ramat (1992) define as Neg + V languages. The main negator is no. It functions as a holophrastic negator with anaphoric function (ex. (1)), but also precedes the finite element in VP (ex. (2)). When the verb governs proclitic pronouns, no is placed before the verb and the pronoun. Here are some examples quoted after Alarcos Llorac (1994: 131). (1) A: Es-tu Juan? Are you John? B: No (2) Juan no cometió el delito Juan neg commit-pa-3s the murder ‘John didn’t commit the murder’ Since Spanish is subject to negation raising (cf. Horn 1989), the particle no precedes the verb whatever its scope. Thus, intonational contours and contextual implications play a crucial role in the interpretation of negative sentences. In total negation (cf. (2) above), the scope of no is the VP. In partial negation, the scope of no (which in our examples below coincides with la violencia) is either to its right (as in (3), an instance of an explicit, cleft, scope-marking construction) or to its left (as in (4)). (3) No es la violencia la que resolverá nuestros problemas neg is the violence it-fem which solve-fut-3s our problems en el verano in the summer ‘It is not violence that shall solve our problems in the summer’ (4) La violencia no resolverá nuestros problemas en el The violence neg solve-fut-3s our problems in the verano (pero la paciencias sì) summer (but the patience yes) ‘Violence shall not solve our problems in the summer’ Other negative items (nunca ‘never/not ever’; nada ‘nothing/not anything’; nadie ‘nobody/not anybody’; ninguno ‘no/not any’; tampoco ‘neither/not either’) can either precede or follow the verb. No co-occurs with post-verbal negative constituents, exhibiting negative concord.



 Patrizia Giuliano

(5) Ese plan no lo aceptará nadie del consejo This plan neg it accept-fut-3s nobody of the council ‘This plan will be accepted by nobody in the council’ Note that several negative words may co-occur in the same sentence as in (6), yet another instance of negative concord. (6) No recibe nunca a nadie neg receive-pr-3s never to nobody ‘He/She never receives anybody’ However, no is not compatible with preverbal negative constituents. (7) Nadie (*no) hará eso Nobody (neg) do-fut-3s that ‘Nobody will do that’ In sum, negation is straightforwardly pre-verbal in Spanish, it is adjacent to the finite element of VP, and it exhibits negative concord.

. Negation in French The main morphemes of negation in French are non and (ne)... pas. Non is a holophrastic negator with anaphoric function, however, it is also used in the negation of verbless clauses. (8) Je prendrai le bus et non le métro I take-fut-1s the bus and neg the subway ‘I shall take the bus not the subway’ To negate verbless clauses, pas may also be used. (9) Italien ou pas, il n’ est pas élégant Italian or neg, he neg is neg elegant ‘Italian or not, he is not elegant’ Through a negative polarisation process, the item pas has become the crucial negator of spoken French,5 which makes familiar French a V + NEG language (cf. Pollock 1989, for a synchronic explication). Because of this development, ne but not pas is normally left out in colloquial French, (as is shown by oral data collected in Paris by François 1974 where 157 tokens of – Ø... pas are found against 3 tokens of ne... pas):

Negation and relational predicates

(10) Je veux pas venir I want-pr-1s not come ‘I don’t want to come’ Pas, but not ne, can be omitted when the utterance contains another negative item (personne: ‘nobody’, rien: ‘nothing’, aucun: ‘no’, plus: ‘no longer/no more’, jamais: ‘never’, ni: ‘nor/neither’ and others). (11) Il n’ y avait personne dans la chambre it not there had-imp-3s nobody in the room ‘There was nobody in the room’ As to syntax, (ne)... pas enclose either the finite verb and its clitic object pronouns, or the (modal/temporal) auxiliary, in a complex VP. (12) Je (ne) lui en ai pas parlé I (neg) him of.it have-pr-1s neg spoken ‘I have not spoken to him about this’ Pas is preposed to the non-finite verb as in (13). (13) Il a peur de (ne) pas s’ en sortir He has fear of (neg) neg himself of.it get out-inf ‘He is afraid of not getting out of it’ As in the case of Spanish negation, the strictly fixed position of (ne)... pas with respect to the VP makes it difficult to identify the scope of negation, at least in written French. These ambiguities are avoided by appealing to a marked intonational pattern (in oral French), or by morpho-syntactic procedures such as clefts (Ce n’est pas au théâtre que je vais [it is not to the theatre that I go]), with or without the adjunction of contrastive continuations (je ne vais pas au théatre mais au cinéma [. . . but to the cinema]) or still the placement of pas – along with non – near the constituent over which the negator has scope (Je vais non pas au théâtre, mais au cinéma). The last strategies are exploited both in spoken and written French. A final observation concerns the use of the partitive de, an element which plays a crucial role in French but not in Spanish, where it is just used in the expression nada de. An utterance such as (14) je n’ai pas de livres I neg have neg part books ‘I don’t have books’ will be translated into Spanish as



 Patrizia Giuliano

(15) no tengo libros neg have bokks As a result, Spanish informants might encounter some difficulty in the segmentation of utterances with partitive where the verb is phonologically weak (il [n’] y a pas de..., il [n’] a pas de... and so on). Spanish informants are faced with a further problem: they must distinguish the anaphoric negator from the verbal negator, a distinction that does not apply in Spanish.

. What needs to be acquired? The acquisition of French negative syntax and morphology raises major problems for Spanish speakers, whose native language has preverbal negation, as we saw in §5.1. In terms of input interpretation, French negative constructions may be difficult to process both because of the postverbal (or discontinuous) character of the negator(s) violating the universal Neg + V, discussed in many typological studies, and of the often obscure character of French phonology.6 As far as transfer is concerned, structures resulting from negative transfer such as the Neg + V strategy, may combine with strategies of a different nature, to support a tendency of wider impact: the Neg + V universal is one of them. A pragmatic and semantic strategy, such as the well known trend to put negation just before the element it affects, could equally explain a preverbal placement of the negator in case the latter has scope over the verb. In other words, transfer phenomena may be extremely difficult to detect and quantify, as the resulting structure is open to multiple interpretations.

. Informants’ learner varieties At the onset of the ESF investigation, Bernarda’s and Palmira’s level of TL proficiency is very low and both of them can be said to be in a prebasic variety. Both of them rapidly progress to a basic variety (BV). In the beginning of their BV, subjectverb agreement marking and tense/aspect references on verbs are completely lacking. By the end of the basic stage some of these features begin to appear on certain verbal categories (be/être and have/avoir) but these occurrences are still rare and the learner’s interlanguage still stays essentially basic. In other words, learners can go through a transitional stage where verb morphology is different in two subgroups depending on the relational or lexical nature of the verb. By the end of the investigation, Bernarda and Palmira have progressed to a learner variety exhibiting postbasic features (PBV). What characterises this final stage is the development of auxiliation and the emergence of a present/past opposition marked on verbs.

Negation and relational predicates 

.. Bernarda A first analysis of the data led us to divide the data collected from Bernarda during the three cycles of investigation into three varieties: the prebasic variety (PreV, namely from the 2nd to the 6th month of stay in France); the basic variety (BV, 13 months of stay in France) and the postbasic variety (PBV from the 24th to 28th month of stay in France). The first negative utterances produced by Bernarda correspond to minimal constructions such as Non + X or X + Non, where X coincides with an item given by the native speaker. (16) Cycle 1 – Encounter 2 (PreV) IN il y a des taxis ‘There are some taxis’ BE non taxis non ‘no taxis no’ Non is equally used as prophrase. (17) Cycle 1 – Encounter 2 (PreV) IN C’est un accident ‘It’s an accident’ BE non + *un* manifestation ‘No + a demonstration’ Bernarda’s first negative utterances with an explicit verb appear in the 5th encounter of the first cycle.7 Here is a passage: (18) Cycle 1 – Encounter 5 (PreV) IN Alors qu’est-ce que tu penses du cours de so what do you-pr-1s think of the course of français? French?’ ‘so what do you think of the French course?’ BE [ke] *nosotros* trois mois [e] un petit [pu] that we three months ?is/eh? a little bit ‘that three months is too short for us’ BE pas [parlar] français not speak-inf French ‘we don’t speak French’ From the basic variety on, the negator par excellence employed by Bernarda is [nepa], a non-segmented structure the informant freely alternates with [nepade]. We will therefore use [nepa(de)] to refer to either of these expressions, and translate them as isanot.8

 Patrizia Giuliano

All along Bernarda’s basic variety, [nepa(de)] plays the role of an “archinegator” since it can appear in any kind of negative context except as a prophrase (a function which is exclusively fulfilled by non) or with the formulae je (ne) sais pas and je (ne) comprends pas, which are always target-like. Given the fixed character of [nepa(de)], it can refer to any person occuring in contexts that demand a copula or avoir (cf. examples (19) and (20) below). This structure always precedes the lexical verb, but follows be, which confirms its non-segmented nature (cf. [nepadekriBir] in ex. (21), and [son + nepa] in ex. (24)).9 Here are some examples: (19) Cycle 2 – Encounter 2 (BV) IN *decile que no estan muy contentos* Tell-im-2s.him that (they) neg are very happy ‘Tell him that they are not very happy’ BE [nepade kontõn] isanot happy ‘they are not happy’ (20) Cycle 2 – Encounter 4 (BV) BE [el demãnd] à la dame *por* [mãZe] ‘he asks the woman for food’ BE *y* [nepade] l’ argent *por* [paZe] And isanot the money for pay-inf ‘and he has got no money to pay’ (21) Cycle 1 – Encounter 7 (BV) BE Mon mari eh [eskri] *y* [kompri] bien le français My husband eh write and understand-pp well the French ‘My husband eh writes and understands well French’ IN Ton mari ? Oui mais ça fait plus longtemps qu’ il est ‘Your husband? yeah but it’ s (a) longer time that he’s là there’ BE Oui oui mais moi [nepadekriBir] Yes yes but me isanot write-inf ‘Yes yes but I cannot write’ As illustrated in (21), [nepa(de)] can be preceded by a topic-subject (moi). Since the copula, be and have are absent from the basic variety, for a long period it is impossible for the analyst to judge the position of the negator with respect to these verbs (cf. (19) and (20) above). From the 3rd cycle on (initial postbasic variety), tokens of être and avoir (both as simple verbs and auxiliaries) are more frequent, which renders the analyst’s task much simpler. Actually, the initial postbasic vari-

Negation and relational predicates 

ety shows two negative syntactic structures: the first one, illustrated in examples (22) and (23), coincides with the French target construction (ne +) Vlex/Mod/Aux + pas, the second one with the basic variety structure [nepa(de)] + Vlex (cf. ex. (21) above). (22) Cycle 3 – Encounter 6 (PBV) BE [il ne pej pa] chômage si la personne [ne he neg pay neg unemployment if the person neg travaj pa] work-pr-s neg ‘They don’t pay unemployment if the person hasn’t worked’ (23) Cycle 3 – Encounter 1 (PBV) BE Mon mari [el kondui] mais [Ze ne pe] pas My husband he drives but I neg can neg ‘My husband he drives but (me) I cannot’ In some passages, the informant combines the morpheme [nepa(de)] of the basic variety with the negative syntax of the postbasic phase (postverbal or discontinuous negation), which results in the structure ne + V + [nepa]. Here is an example: (24) Cycle 3 – Encounter 6 (PBV) – Spatial Description IN En bas? Où ça + en bas?. . . Down? where (is) that + down?. . . BE *como* [s apel] les choses quand [son + nepa] sur la How pro call the things when are + isanot on the tabla? table? ‘How do you call things when they are + not on the table?’ IN Sous? (under?) BE [su] (under) In the postbasic variety, negation always follows être and avoir but either precedes or follows the lexical verb. This distribution is not arbitrary, and the following principle applies: In order to precede pas (or [nepa]) the verb must be finite or relatively finite “Relatively” refers to the fact that the (thematic or non-thematic) verb does not necessarily convey all the information associated with the notion of finiteness (tense, aspect, subject-verb agreement) but just part of it, especially tense (present

 Patrizia Giuliano

vs. past, expressed by Vstem and Aux + Vlex, respectively) and/or subject-verb agreement (cf. Giuliano, in press). Aspectual marking is much more ambiguous; a clear instance of aspectualised morphemes is given by the very few occurrences of French futur simple and futur proche. Table 1 contains all negative utterances with an explicit verb produced by Bernarda during the data collection period; the formulaic expressions Je (ne) sais pas and Je (ne) comprends pas have not been included: Table 1. Verb Negation in Bernarda: preverbal [nepa(de)] as fixed form; postverbal [nepa] and (ne/[ne])... pas

[ne(pade)] + Vlex [na]/ne + Vlex + pas (no +) Cop + pas + X ne + Mod + pas (+ Vlex) (ne +) Vlex/Cop + [nepa] + beaucoup/bien Aux + [nepa] + Part. Passé Presentationals [senepa; ilnjapa...] Total

PreV

BV

PBV

1 – – – – – – 1

19 – – – – – – 19

26 57 5 4 5 1 21 119

.. Palmira Palmira’s data can also be divided into three varieties. The prebasic variety (the first three months of data collection), the basic variety (extending over about a year) and an initial postbasic variety (the final 6 months of observation). On the whole, Palmira produces less negative utterances than Bernarda and uses the construction no + V more frequently. Nevertheless, during the basic variety stage, Palmira begins to produce negative utterances with a negator similar to the one used by Bernarda, the non-segmented morpheme [nepa], which in Palmira, too, freely alternates with [nepade]. Here there are some extracts from Palmira’s BV: (25) Cycle 2 – Encounter 2 (BV) PA *y un* monsieur *como se dice delgado un* And a man how pro10 say-pr-3s slim? A monsieur [ke] *no* [nepade gorS] + *en el* bateau de la man which no isanot big + in the boat on the droit right ‘And a man how do you say slim? A man who is not fat + in the boat on the right’

Negation and relational predicates

(26) Cycle 2 – Encounter 2 (BV) IN ah d’accord et c’est long le voyage? ‘Ok and it’s long the journey?’ PA eh *por* hm oui *por* Milton eh eh for hm yes for Milton eh PA [nepade konose] la / [kono/ konos / konose] ? isanot know the / know/knows/knowing?11 IN il ne connaît pas la he neg knows neg the ‘he does not know the’ PA la route the trajectory As the examples above show, [nepa(de)] may appear together with an explicit (26) or a non-explicit (25) verb, where the TL would generally have avoir or être. The negator [nepa(de)] is never used as a prophrase, so it never replaces non in this function. From the (initial) postbasic variety on, Palmira uses target-like negative structures with postverbal or discontinuous negation. Preverbal negation with [nepa(de)] does not disappear but it is used in just a few instances with formally ambiguous lexical verbs. Conversely, with finite copula, avoir and auxiliaries, the informant uses only postverbal pas. By this last period of data collection, the learner clearly marks (although not yet systematically) some functional oppositions on the verb: the perfective (through the Aux + Vlex construction), the prospective (by French futur simple) and the imperfective of être, by the morpheme [ete]. Table 2 shows all negative utterances with an explicit verb produced by Palmira throughout the data collection period; the formulaic expressions je (ne) sais pas and je (ne) comprends pas have not been included. Table 2. Verb Negation in Palmira preverbal [nepa(de)]; (ne/[ne])... pas

[nepa(de)] + Vlex (ne) + Vlex + pas (ne) + Cop/Avoir + pas + X/Vlex (ne) + Aux/Modal + pas + Vlex [senepa] Total

PreV

BV

PBV

– – – – – –

4 – – – – 4

3 22 4 5 12 46



 Patrizia Giuliano

. From Italian to English . Negation in Italian The Italian system of negation is in many respects comparable to that of Spanish, since it involves preverbal negation by means of the morpheme non (cf. the analysis of Zanuttini 1997). This operator immediately precedes the (lexical, auxiliary or modal) verb except with clitics. Here are some examples. (27) Mario non ha visto Nina Mario not has seen Nina ‘Mario hasn’t seen Nina’ (28) Mario non l’ ha vista Mario not her has seen ‘Mario hasn’t seen her’ Under specific pragmatic conditions, non can move next to the argument it affects (including the syntactic subject) or in the focussed part of a cleft, along with the element in its scope. Most of the following examples come from Manzotti and Rigamonti (1995: 250–252), and Bernini and Ramat (1992: 100–101): (29) Giovanni mangia non pesce ma carne ‘John eats not fish but meat’ (30) Non Giovanni ma Maria mangia pesce ‘Not John but Mary eats fish’ (31) Non é Giovanni che mangia pesce ma Maria Not is John who eats fish but Mary ‘It is not John who eats fish but Mary’ Example (29) can be replaced by a sentence with a preverbal negator (Giovanni non mangia pesce ma carne). In this case, differences of scope will be signalled by different intonational patterns. The syntactic interaction between non and other negative elements gives rise to two situations: a.

non precedes VP whereas the other negative elements (nessuno/alcuno: ‘nobody’; mai/giammai: ‘never’; niente/nulla: ‘nothing’; più: ‘never again’; affatto: ‘not at all’; neanche/neppure/nemmeno: ‘not... either’; né... né: ‘nor... neither’) follow it; b. the other negative element precedes the verb, which causes non to disappear and gives rise to an emphatical interpretation of the utterance. Here are some examples:

Negation and relational predicates

(32) Non è venuto nessuno/alcuno Not is come nobody ‘Nobody has come’ (33) Nessuno è venuto Nobody is come ‘Nobody has come’ It is also possible to combine several negators in the same clause. (34) Nessuno dirà mai niente ‘Nobody say-fut-3s never nothing = Nobody will ever say anything’ Although Italian belongs to the group of Romance languages that Bernini and Ramat (1992) define as Neg + V languages, spoken Italian also presents a negative construction with discontinuous negator(s), namely non... mica. It is a pragmatically marked structure (strong refusal), as was also the case, in the past, for the postverbal or discontinuous negator of languages such as French. Here is an example (cf. Bernini 1992: 191). (35) A: B:

Hai capito? ‘Have-pr-2s (you) understood?’ Certo che ho capito. Non sono mica stupida ‘Of course that have-pr-1s understood. Neg am neg stupid io! me! ‘Of course I have I am not stupid!’

The meaning of non... mica is to refuse the (explicit or contextually deductible) presupposition of an assertion and not the assertion itself. The sentence non sono mica stupida, in example (35), is the answer to hai capito? which presupposes something like ‘It’s possible that you haven’t understood’. In questions, mica performs the function of repeating the speaker’s negative expectation conveyed by preverbal non (cf. ex. (36)), or that of conveying a courtesy nuance, given that the listener can answer in a negative way to the request (cf. ex. (37)). The examples below are from Manzotti and Rigamonti (1995: 284). (36) Non é mica arrivata Maria? Neg is neg arrived Mary? ‘Mary hasn’t arrived yet, has she?’ (37) Non hai mica una sigaretta? Neg have-pr-2s neg a cigarette? ‘Would you have a cigarette?’



 Patrizia Giuliano

As to prophrases, Italian disposes of sì ‘yes’ et no ‘no’. Here are some examples with no. (38) A:

B:

Vuoi ballare? Want-pr-2s danse-inf? ‘Do you want to dance?’ No / Con te no ‘No / with you no’

No is also used with a large range of discourse functions (self-correction, surprise reaction, monitoring on the reception of the message from the listener...),12 as it is also the case, at least for some functions, for Spanish, English and French prophrases.

. Negation in English English negation is in contrast to the most widespread type of negation in Europe. The analysis of Pollock (1989) convincingly shows the strong relation the negator not entertains with the finite component of VP, which justifies raising of negation over the lexical verb – inflectionally weak in English – to INFL (do, does, did). Negation being sensitive to INFL, Pollock’s analysis justifies the fact that not always follows auxiliaries and copula13 which because of their inflectionally richer morphology can raise to INFL – and systematically precedes infinitive and gerund (cf. (41)), namely the non finite component of VP. Here are some examples: in the first and the second ones not is contracted into the very frequent colloquial morpheme n’t (don’t, doesn’t, isn’t, won’t...): (39) He doesn’t go to the cinema (40) He isn’t going to the cinema (but to the theatre) (41) He told me not to follow him This particular functioning of negation makes English a language with postfinite rather than postverbal negation. A further aspect separates English from the other languages involved in our study: it is the only European language which presents a clear distinction between negative quantifiers with presupposition of non-existence (nobody/no one, nothing, nowhere, never), neutral quantifiers with presupposition of non-specific existence (anybody/anyone, anything, anywhere, ever) and positive quantifiers with presupposition of specific existence (somebody/someone, something, somewhere), which rarely overlap. This pattern differs from Italian, French and Spanish where there is overlap between neutral and positive quantifiers. In agreement with this tripartition of quantifiers, English equally has three classes of indefinite adjectives: no, any

Negation and relational predicates

and some. These adjectives allow English speakers to mark negation either on the verb (ex. (42a)) or on the NP (42b). (42) a. b.

I haven’t any money I have no money

In both examples the negator affects the assertion itself but with different pragmatic nuances, since “... the combination of not (n’t), and the non-assertive word is more colloquial and idiomatic than the negative variant” (Quirk et al. 1985: 377). This particular aspect of English involves three considerations: a. English is forced to signal the absence of reference of NP; b. point (a) justifies nominal constructions such as no idea, which are not permitted in Italian or Spanish; c. points (a) and (b) are pertinent for all Germanic languages, which suggests the hypothesis that in these languages negation be directly tied up to the focus of utterance, namely to pragmatic rules. Together with the “post-finite” character of negation, points (a), (b) and (c) equally differentiate English from Italian.14 A last difference between these two languages lies in the fact that English does not permit the combination of several negators in the same utterance (cf. (43)) unless we want it to be positive (cf. (44)): (43) Nobody ever says anything for any reason15 (44) I can’t not obey (= I have to obey) But English and Italian negation do also share some properties. Both languages have the same negator for the prophrase function (no) and a similar negator for sentence negation (Eng. not vs It. non). Furthermore, in restricted scope contexts English allows options comparable to those commented for Italian: if the negator keeps its position between the auxiliary (do, temporal or modal) and the lexical verb, the interpretation of the utterance will crucially depend on the intonational pattern; alternatively, the negator can be moved beside the constituent it has scope over, which causes the do auxiliary to disappear, as in (45). (45) He hit not John but Robert As we can see, (45) necessarily demands a contrastive continuation (... but Robert). It is equally possible to opt – just as in Italian – for a third alternative, namely a cleft construction: (46) It’s not John who was hit (but Robert)



 Patrizia Giuliano

. What needs to be acquired? Italian speakers learning English have to tackle three main problems: a.

they must realise that forms such as don’t, doesn’t, didn’t, shan’t, won’t... are not simple negators but negative auxiliated structures; b. they must capture the “postfinite” character of the negator not (and its contracted version, n’t), which means becoming aware of the fact that the negator follows the copula and modal or temporal auxiliaries but also ties up to a “dummy auxiliary” (do, does, did) under certain circumstances; c. they must capture the cancellation of not with negative quantifiers (cf. also below) and the obligation, in English, to signal the non-reference of an NP by neutral or negative indefinite terms. Points (a) and (b) clearly show that the mastering of English negation involves the mastering of finiteness. Now, for a learner (Italian or not), this correlation can be extremely difficult to discover, because if the postfinite placement of not/n’t is phonetically only relatively evident with be, have and modal auxiliaries (can, may, need...), it is certainly less clear with do, does, did, shall, will, etc. Both the syntax and the morpho-phonology of English negation can complicate the acquisitional task for the Italian learners. As for syntax, two learner strategies are to be envisaged: 1. because of the high frequency of the structure don’t + lexical verb in the input and the possible interpretation of don’t as a simple negator, learners will favour the preverbal placement of negation with any type of verb (lexical, modal, auxiliary or copula). L1 negative syntax and information structure could reinforce this hypothesis; 2. the postfinite character of not/n’t with some predicates leads informants to over-generalize this rule with any verb category (copula, auxiliary or lexical). As to the morphology of negation, in English the sentence negator (not) has a morphological and phonological realisation very similar to that of the anaphoric negator (no) and the quantifier no: this fact could lead learners to deviant morphological hypotheses with respect to the TL, in particular, to an over-generalisation of no. This would confirm a tendency identified in other studies on English as L2 (cf. Cancino, Rosansky, & Schumann 1978; Stauble 1978) and on languages that have very similar morphemes for the prophrase and the sentence negator (cf. Bernini 1996 and 2000 on the acquisition of Italian).16 We should also comment on point (c) of the beginning of the paragraph. The cancellation of not with a negative quantifier is a normative phenomenon of formal English but not of the colloquial language, where it is possible to use several negators, because of the greater functional weight of negative items (Nobody gave him nothing / Nobody didn’t give him nothing = Nobody gave him anything).

Negation and relational predicates

From a learner’s perspective, strongly marking the negative force of the utterance he has to produce may help to make his message clearer to his listener, especially when his mother tongue allows constructions with multiple negation, as is the case for Italian.

. Informants’ learner varieties At the onset of the ESF investigation, both Andrea and Santo were in the basic variety stage. Nevertheless, some very rare occurrences of relational be and have are already attested. By the end of the longitudinal study, only Andrea had progressed to a postbasic variety.

.. Andrea At the time of the first encounter, Andrea had been in England for approximately eight months. On the whole, Andrea’s data show an evolution from a basic to a relatively postbasic variety: an essentially basic stage from the 1st to the 7th encounter of the 1st cycle (8 months of observation); a very initial postbasic variety, from the 8th encounter of the 1st cycle up to the 7th of the 2nd cycle (this time span corresponds to 7 1/2 months of investigation), and a relatively more advanced variety, from the 8th encounter of the 2nd cycle until the end of the 3rd cycle (8 months of observation).17 During the first part of the inquiry (1st cycle), the informant already uses is as copula, beside a very high percentage lexical verbs, which are marked neither for tense norfor grammatical aspect. By the end of the 2nd cycle, Andrea uses complex verb groups, which proves his initial ability to identify certain finite morphological elements as opposed to non-finite elements (for example is possible to send?, I want to send, I have to decide, I have to ask) and also alternates is/am/are with has/have. Actually, be and have are particularly sensitive to finiteness and the structure be/have + lexical verb is frequently used to refer to perfect or perfective past. Instances of regular past tense are not lacking (for example this girl burgled and one policeman seed her = TL ... saw her), and are certainly more reliable than occurrences of irregular past tense, which are often used in contexts referring to the present.18 From the 8th encounter of the 2nd cycle, Andrea keeps on elaborating the morphological distinctions that have evolved in the 7 preceding months, in particular, the present/past opposition, which certainly shows development although his interlanguage is still that of an initial postbasic variety.19 As to the expression of negative functions, on the onset of the investigation, Andrea often applies an over-generalised negator, the morpheme no. This negator is employed in different contexts: No + N/Adv with anaphoric value; No + V; NP + No + Adj with copula ellipsis; no as prophrase. Here are some extracts:



 Patrizia Giuliano

(47) 1st Cycle Encounter 3 (BV) AN People out the room thinks er ‘People outside the room think er one person in the room. . . there’s someone in the room’ AN no person in the room ‘There’s no person in the room’ IN There’s no person in the room (48) 1st Cycle Encounter 1 (BV) IN You can’t get that job in England?. . . AN I prove two time but er + no speaking ‘I tried twice but er + I don’t speak’ AN No speak English ‘I don’t speak English’ The structure no + Vlex is at this stage, the most frequent negative construction. But, strikingly, Andrea never exploits preverbal no with the copula, i.e. the only non-thematic verb he disposes of in the BV. At the BV stage, Andrea also uses the negators don’t and not (or its contracted form n’t). Don’t is almost exclusively employed with the formula I don’t know; a few other lexical verbs can be found with don’t; (49) 1st Cycle Encounter 1 (BV) IN Where’s the tunic? AN I don’t know (50) 1st Cycle Encounter 3 (BV) IN What other things could you see in the room? AN I don’t see very well ‘I couldn’t see very well’ As for the negator not, it freely alternates with n’t; the latter is the most recurrent variant. At the BV level, both variants exclusively appear with copula, always in postverbal position. (51) 1st Cycle Encounter 1 (BV) AN is not shoes? ‘aren’t they shoes?’ In the PBV, negative utterances show signals of syntactisation: no/not/n’t are always placed after the non-thematic verbs be, have and can; don’t systematically appears with lexical verbs, and the structure no + Vlex has almost disappeared (cf. PBV1 in

Negation and relational predicates 

Table 3. Verb negation in Andrea: Negative utterances with an explicit verb

No + Vlex No possible + Vlex don’t + Vlex is no/not/n’t + X20 is no possible; is impossible haven’t + X can’t /cannot + (to +) Vlex Wouldn’t Total

BV

PBV1

PBV2

13 2 9 3 – – – – 27

2 – 13 13 3 4 2 – 37

– – 12 3 – 7 6 1 29

Table 3). The end of the investigation marks the complete disappearance of preverbal no (cf. PBV2 in Table 3). The modal expression no possible, used before, equally disappears and is replaced by impossible or can’t. Nevertheless, Andrea will never inflect the auxiliary do with respect to the 3rd singular person of present tense (cf. target English does) and past tense (cf. did). The table above sums up Andrea’s acquisitional path of English negative morpho-syntax.

.. Santo The transcription of Santo’s 3rd cycle of recordings is not available at the ESF data bank. The last transcribed recording available is from the 4th encounter of the 2nd cycle, corresponding to 20 months of his stay in England and 13 months of investigation. When the ESF data collection began, Santo had been in England for 9 months. Santo’s interlanguage never really developed beyond the basic variety. As far as negative particles are concerned, their acquisition by Santo is slow and incomplete. Nevertheless, he maximally exploits the means he disposes of by continuously experimenting with different structural combinations which allows him to express relatively difficult negative concepts. The negator the informant mostly uses is no, which covers any negative discursive function (cf. the overgeneralization of no in Andrea) and which also helps to create antonyms (cf. no happy, “unhappy”, in the 6th encounter of the 1st cycle21 ). The same negator frequently appears in the No + X pattern. No + X is frequently used in interactions with interviewers, where the learner can exploit the information his listener has already introduced in the conversation, as illustrated in (52); (52) 1st Cycle Encounter 1 IN Did you learn some French at school? SA no no IN No? SA No French

 Patrizia Giuliano

IN Just some English? SA yeah Contexts where no goes along with two or several constituents (which precede, follow, or both precede and follow the negator) normally correspond to structures which are not dependent on the discourse produced by the native speaker. In these constructions the learner juxtaposes items without any connector or verb. This second type of structure is plausibly more complex. Here is a passage: (53) 1st Cycle Encounter 2 SA I very *agitato* today. . . ‘I am very nervous today. . . ’ SA for me no very + concentration22 ‘I don’t have a lot of concentration’ Despite the over-generalised use of no, Santo will never use it with the verb know, because – evidently – of the formulaic character of the expression I don’t know. All along the period of observation, don’t is never inflected for 3rd person singular (doesn’t), neither for the past tense (didn’t). It can be combined with nonproductive verb items with -ing suffix or irregular past tensed verbs. As to other negative items, and in particular the quantifiers, Santo’s interlanguage presents several passages where these elements play, sometimes, idiosyncratic functions with respect to the TL: nothing (= TL nothing, nobody, no one), no anything (= TL not at all), never (= TL never), never mind (= TL never mind, doesn’t matter), no everything (= TL not all of them, not any, nothing). In the following examples Santo uses no everything and never mind. (54) 1st Cycle Encounter 6 (The use of fireworks in Italy) IN122 but it’s not within the law you said. It’s against the law? SA against the law IN1 ya, is it? + You translate? IN2 ya. E’ contro la legge?. . . SA no everything *contro la legge* ‘Not all of them are against the law’ SA *se è* big one *fa* bum bum big one *allora* possible *contro la legge* ‘If it is big it makes bum bum it’s big then it probably is against the law’ SA no possible for State. . . ‘the State interdicts them’ SA for State no possible this fire ‘The State interdicts this kind of firework’

Negation and relational predicates

(55) 2nd Cycle Encounter 2 LN you like watching sport yeah SA yeah because is never mind for understanding language, you know? ‘Yes because it’s not important to understand the language, you know?’ The informant also produces utterances with double negation as, for example, don’t/no... nothing: (56) 1st Cycle Encounter 9 SA manager + this house ‘The landlord of this house’ IN ya SA have the another one house *no*? ‘He’s got another/a second house right?’ IN ya ya SA in Castle Road IN ya i know SA and live + two years one person *e* no pay nothing ‘And a person has lived there two years and didn’t pay anything’ The absence of any modal verb in Santo’s data can be explained by the fact that he exploits alternative modal markings, namely the locution no possible, which Andrea equally produces although in a much more limited way (Andrea extensively uses can; cf. Table 3 in §6.4.1). This protomodal function is confirmed by the observation that possible can appear as a modalising expression in positive utterances, too (e.g. for six months this possible drive [= TL I can drive with this for six months] and possible speak English [I will be able to speak English]).24 Table 4. Verb negation in Santo Negative utterances with an explicit verb

BV

No/non/not + Vlex no + have have/Vlex + no/not + X don’t + Vlex Cop + No/Not/No possible (+ X) Total

93 2 2 49 46 192

Table 4 gives an overall view of negative utterances. The analysis of this table shows that there are both changes and stable elements during the basic variety. Other



 Patrizia Giuliano

analyses (Guliano in press) show however that there is some development which can be summarised as follows: – – – – –

the appearance of the copula in negative utterances; the postverbal position of the negator with respect to copula; the massive exploitation of not; the increasing frequency of don’t with non-formulaic negative utterances containing a lexical verb; the appearance of new negators (never, nothing, nobody...) with an explicit verb.

The stable elements are essentially: – – – –

the massive exploitation of the structure no/not + V lexical; the high number of (positive and negative) nominal utterances; the persistence of the protomodal expression no possible and the absence of any other modal; the non-productive character of verb morphology.

The comparison of new and stable elements leads to a result quite different from that observed in Andrea’s final acquisitional stage. Postverbal negation is exploited just with the rare occurrences of copula, i.e. the only inflected verbs Santo produces. The over-generalised function of no does not dissappear. The new phenomena observed do not result in a real reorganisation of Santo’s interlanguage, which still shows the organising principles of a basic variety that just has a larger lexical repertoire and more complex utterances.

. A common acquisitional path The analysis of our Spanish and Italian informants presented above lead to the identification of some common stages in the acquisition of negation. Learners follow an acquisitional path consisting of essentially three stages, which correspond to the stages identified by Klein and Perdue (1992): the nominal stage (or prebasic variety), the non-finite verb stage (or basic variety), and the finite verb stage (or postbasic variety). In each phase there are specific features in the way the informants mark negation and in the relation between negation and verbal morphology from the late basic variety on. The following schema summarises the common trends. It simplifies reality by disregarding the fluctuating character of each phase which we observed in the detailed analysis of the informants. Schema 1. Acquisitional Path of Negators Prebasic Variety (NP) + Neg + X X = N/Adj/Adv Bernarda 1st Cycle

Negation and relational predicates 

IN C’est un peu une imprimerie alors ? ‘It’s a bit a printing then?’ BE non non non [nepad eprimeri] *otra forma* no no no isanot printing other form ‘No no no it’s not a printing other thing’ Andrea 1st Cycle IN I see other system + *agopuntura*. . . ‘I saw another system + acupunture’ IN it’s only here (with a hand gesture) AN No inside ‘not inside’ Basic Variety (NP) + Neg + X (+Y) X= Vlex non-finite Y= Indirect Object Bernarda 2nd Cycle IN Quand tu parleras très très bien le français tu te sentiras mieux je crois ‘When you speak French very very well you will feel better I think’ BE oui oui oui [ZE krwa] hein + [paske] très très difficile Yes yes yes I think hein + because very very difficult [nepa parle] bien le *frances* isanot speak well the French ‘Yes yes yes I think + because very very difficult I don’t speak French well’ Andrea 1st Cycle IN And you can’t get that job in England? AN i prove two time ‘I tried twice’ AN but er + no speak English no ‘but er + I don’t speak English no’ Postbasic variety NP + Aux/Vlex/Mod + Neg + Vnon-finite + Y. . . Bernarda 3rd Cycle IN Et il est mort là bas ou il est sorti ? ‘And he has dead there ou he has come away?’ BE Non *el* [e ] là bas [ne pe pa sortir] eh parce que ‘Non he is down there (he) cannot come away eh because [e] dans la prison (he) is in prison’ Andrea 3rd Cycle AN I wouldn’t take

 Patrizia Giuliano

At the beginning of the acquisitional path, our learners exploit a pattern which is extremely productive: the pattern (SN +) Neg + X. X corresponds to a noun, an adjective or an adverb that is in the scope of negation. At that stage, a transparent relation is established between the negative operator and its scope. From the basic variety on, when verbs begin to appear, the learners (Italian and Spanish) exploit a preverbal negation strategy ([nepa(de)] + V; no + V); verbs are uninflected (they lack finite features) and they always correspond to lexical verbs, since être/be and have/avoir are normally lacking. This changes after the morphologisation of the verb (postbasic stage). With the development of verb inflection, the preverbal negation strategy is progressively abandoned and replaced by postverbal/postauxiliary negation or discontinuous negation ((ne) + V + pas; Aux/Cop + not; don’t/doesn’t + V). The stepwise acquisition of verbal morphology is summarised in Table 5. Initially, verbal inflection appears exclusively with some forms of the copula (be/être) and avoir. Some time later, the first occurrences of temporal auxiliaries appear. The copula and avoir as well as auxiliaries appear according to the principle singular before plural. As for English, has appears later than have; nevertheless, has precedes any other 3rd singular person present tense (cf. Andrea, who exploits has but who doesn’t master the ending -(e)s with lexical verbs). The very first (and still very rare) occurrences of être/be and have/avoir appear at BV level. This statement has nothing contradictory if we admit that the passage from one stage to another can be extremely slow, above all when we observe such a delicate phenomenon as verb morphology (cf. Bartning 1997). For informants who progress to a postbasic variety, the mastering of être/be and avoir/have is followed, some time later, by that of modals. The Spanish informants learning French, also use some inflected lexical verbs that were not included in the tables for reason of simplicity. Bernarda, for example, acquired [di] = TL dis/dit; [va] = TL va; [sor] = TL sors; [fe] = TL fais/fait; [apran] = TL apprends/apprend. Whereas modals just appear in the singular, lexical verbs follow the same principle as être and avoir: singular before plural. Auxiliated structures precede suffixed morphemes referring to past or future (cf. the late appearing of V-ed in Andrea and that of imparfait in Bernarda and Palmira) and morphological alternations (buy – bought, catch – caught...) because of their more transparent nature.25 The first modal to appear is can/pouvoir, which in the target languages under investigation emerges after constructions with an auxiliary. For three of our four informants, transfer from L1 cannot explain the preference for auxiliated structures. Spanish speakers, because of their South-American origin, use both perfect with auxiliated structures (pretérito perfecto compuesto), and perfective with suffixed non-auxiliated forms (pretérito perfecto simple) in their L1. The same functional opposition is quite productive for Southern Italian (spo-

Negation and relational predicates

Table 5. Verb Morphology Evolution26

ken by Santo, but not Andrea), where passato prossimo is in strong competition with passato remoto. As a result, the precocious appearance of auxiliated structures cannot be interpreted as a reflex of aspectual and temporal codifications of the L1. Some of these Aux + Vlex constructions are equally used to imperfective actions in the past. To sum up, the emergence of functional morphology involves the following implicational scale: Schema 2. Be/être – Avoir/Have > Aux + Vlex > Modals/Irregular Verbs/Suf fixed Forms



 Patrizia Giuliano

Table 6. Bernarda27

. Negation and finiteness In this section we shall try to highlight the relation between the evolution of verb morphology and the different ways of expressing sentence negation along the acquisitional path. Tables 6–9 in this section below sum up the morphology and syntax of negators and the relevant properties of verb morphology in positive utterances (cf. 2nd column) in the four informants. Through a comparison of these two domains, general trends can be identified by correlating the morphological progression illustrated in the 2nd column of each table to the progression shown in the other columns, concerned with negative structures: structures with preverbal negation and lexical verbs in the 1st column; structures with postverbal negation and lexical verbs, copula and (temporal or modal) auxiliaries, respectively, in the 3rd, 4th and 5th columns. The vertical columns illustrate the morpho-syntactic progression of verbs and negation at any moment of the acquisitional path. In a horizontal perspective, the tables show the

Negation and relational predicates 

Table 7. Palmira

correlations between the acquisition of verbal morphology and the morpho-syntax of negative utterances. Tables 6 to 9 demonstrate that if the verb shows signals of morphologisation, there is always a progression in the domain of negation, too. For the less advanced informants in the beginning of the data collection (cf. Bernarda and Palmira), the postverbal position of the negator (columns 3, 4 and 5 of Tables 6 and 7) only appears when the learner begins to exploit certain morphological alternations on verbs (cf. 2nd columns) in a productive way along the 3rd cycle (PBV). For initially basic learners (Andrea and Santo) postfinite negation equally emerges in correlation with the development of verbal morphology. As the informants exploit certain morphological oppositions on verbs, preverbal negation decreases and over-generalised negators disappear (cf. disappearance of No + V in Andrea, Table 8, column 1).30 Furthermore, postverbal negation increases and is exploited in complex verb groups (cf. columns 5 in Tables 6–8 for Bernarda, Palmira and Andrea). Santo, whose variety fossilises at a basic level, employs non-finite verbs – except for occurrences of the copula and one auxiliated construction with is – and keeps exploiting the No + V strategy in a massive way. The only – rare – signs of finiteness are marked on be.

 Patrizia Giuliano

Table 8. Andrea31

In other terms, the learner’s massive exploitation of the preverbal negation strategy reflects the lack of functional morphology. This strategy is given up whenever informants progress remarkably in the domain of verbal morphology. In this case, finite (or relatively finite) relational predicates always go along with postverbal negation. For a long period, this progression can go along with some of the “old” strategies, which is completely in agreement with the dynamic character of interlanguages (cf. the survival of pas/[nepa] + Vlex in Bernarda and Palmira, and that of fixed don’t in Andrea). In this case, the preverbal negator normally goes along with non-finite lexical verbs. The studies of Bardel (2000) and Schlyter (1998a, b), concerned with Italian and French as L2s, seem to confirm our results. Verbs with poor or no lexical content as Italian essere ‘be’ and avere ‘have’ (but not modals) appear with postverbal negation. In Bardel’s data these same verbs are the first ones to be inflected. This is the more surprising since Italian does not demand a postverbal negator. In her Swedish learners of French, Schlyter observes a precocious appearance of productive inflection on the copula, auxiliaries and modals (so called CAM predicates) and an equally precocious exploitation of postverbal pas with these

Negation and relational predicates 

Table 9. Santo

verbs. In the author’s terminology, this could mean that learners first capture the notion of finiteness and functional categories by CAM predicates, and that they only later realise which function to assign to which form for lexical verbs. These observations, together with the interpretation of our implicational pattern 2 (see schema in Section 7) raise the question of what “functional verb morphology” means in a learner’s variety. The answer lies in the fact that in not very advanced stages verb morphology can be disfunctional in the sense of the target language but functional in the sense of the learner’s logic: in other words, as often happens in learner varieties, phenomena that do not make sense for the native speaker of the TL in question, correspond to specific hypotheses formulated by the learner. Structures with an auxiliary, for instance, mark a remarkable progress with respect to the basic stage but the learner still can be far or relatively far from capturing the temporal and aspectual distinctions encoded by the TL constructions. To put it more precisely, the learner chooses among these constructions those which most closely correspond to his needs at a certain stage.

But why auxiliaries? The precocious finite character of relational verbs could in fact be favoured by their more transparent functional morphology. Modals (in particular, may/can and pouvoir) may appear later since learners can exploit alternative modal expressions (e.g.

 Patrizia Giuliano

[nepa] possible / [se] possible and possible / no possible). Yet, we have to answer a crucial question: Why should postverbal negation be so sensitive to relational predicates? For this point we shall propose a pragmatic explanation. Relational predicates appear in a specific informational configuration: they are situated at the topic-focus barrier and mark the Topic Time, namely the time span for which the utterance makes an assertion (cf. Klein 1994).34 Now, with respect to negation, we have observed that: 1. the negator always follows avoir/have and être/be, which are inflected; 2. whereas the use of inflected forms of have/avoir and be/être is relatively precocious, lexical verbs show greater morphological variation until the first postbasic stages, and the position of the negator is equally variable with respect to such verbs. The observation that être/be and avoir/have are systematically situated on the topicfocus barrier and mark the Topic Time has an extremely important implication for negation: the negator cannot have scope over the Topic Time because the nonvalidity of a propositional content must necessarily be asserted for a given time span. As a consequence, the negator follows be/être and have/avoir, i.e. the elements which carry topic time information, from early stages onwards, but precedes, in that same phase, the lexical verbs on which temporal and aspectual markings are lacking or ambiguous. In other words, the lexical verb without clear Topic Time reference follows the negator and is IN its scope. We can summarise these observations as follows: Verb forms which are clearly non-inflected never precede the negator. Forms preceding the negator may show some morphological variation, but never correspond to true infinitives or past participles: our learners of French never produce utterances such as, *il [sortir] pas even though they produce such sentences as, il [nepa]/pas [sortir] For English L2, we observe the absence of finite lexical verbs after or before the simple negator not, *he not goes he goes not or after don’t as non-segmented negator, *he don’t goes

Negation and relational predicates

The only lexical forms following not or unanalyzed don’t are basic forms as, he don’t/no/not go In conclusion, the implementation of postverbal negation demands a remarkable reflection on the morphology and syntax of the TL, at least for languages such as English and French. In particular, we propose to situate the hypothesis of the functional nature of verbs in a larger perspective than the strictly syntactic approach adopted by Parodi (2000), according to which functional verbs would be simple protomarks of subject-verb agreement. In our opinion, they prototypically convey information such as tense and/or aspect, which do not have a syntactic but rather a functional nature.35 With respect to the postverbal/postauxiliary negation pattern, a possible transfer from the source languages (Italian and Spanish) cannot justify its use since, both Spanish and Italian, follow the rule ‘Neg before the verb’ whatever its nature (lexical or relational). A pragmatic strategy (Neg + X) cannot explain the postverbal/postauxiliary placement of the negator in our data either, because, depending on the context, the scope of NEG can be the focus, the contrastive topic or the whole assertion relation. These observations may shed new light on the differences between first and second language acquisition. The acquisition of verb morphology certainly plays a role in leading our second language learners to an appropriate employment of negators. Nevertheless, that happens because in French and in English verb inflection is strictly connected to negation. So, whereas in children acquiring the mother tongue, verb inflection always plays a role with respect to the employment of negation, even when the language to learn is Italian (a language in which the use of negation does not strictly involve the mastering of verb inflection), the situation seems different in adult learners of a second language. On the basis of Schlyter’s (1998, 1999), Parodi’s (2000), Bardel’s (2000) and our own results, a further difference between children and adults lies in the fact that adults do not learn verb inflection simultaneously with lexical verbs and relational verbs as children do. Adults seem in fact to exploit a less abstract strategy: the inflection of relational verbs precedes that of lexical verbs, since the former have a more transparent morphology. This same idea leads learners to privilege auxiliated structures in order to divide the functions to convey: Aux marks tense and/or aspect and/or agreement whereas Vlex carries the lexical information.





Patrizia Giuliano

Symbols and terminology S M F INF IM PA PaRé IMP PL FUT PR PP Fut Sim Fut Pr PRO 1 3 PART X / it. fr. engl. ?....? [. . . ] (. . . ) Vlex Aux Mod Cop PreV BV PBV TL scope

= singular = masculine = feminine = infinitive = imperative = perfective past = passé récent = imperfect = plural = future = present tense = past participle = futur simple = futur proche = pronoun = 1st singular person = 3rd singular person = partitive = variable (see Note 19) = auto-interruption = Italian = French = English = encloses a word whose interpretation is doubtful = encloses phonetic transcription = encloses a word/words added by the analyst = lexical verb = temporal auxiliary = modal auxiliary (generally, pouvoir and can) = copula = prebasic variety = basic variety = post-basic variety = target language = the actual element/s affected by the negator

Negation and relational predicates

Notes . In our terminology, “sentence negation” refers to negative utterances containing a verb and not to the scope of the negator, which can affect either the assertion or a single constituent. . As several studies on the basic variety have pointed out (cf., for instance, Noyau et al. 1995; and Starren 2001), the basic stage does not necessarily lack tense or aspect markings although such information is conveyed by means other than verbal morphology, namely temporal and aspectual adverbs. . Authors such as Bartning (1997), Benazzo (2000) and Kihlstedt (1999) also describe the tortuous path leading relatively advanced learners to a target use of L2 verb mophology. . Cf. Clahsen, Meisel, & Pienemann (1983). . The word pas comes from latin passum – namely, from a nominal element – which, initially, was probably used exclusively with movement verbs. . Cf., for instance, Dahl (1976) and Bernini and Ramat (1992). . We have not taken into consideration the “formulae” je (ne) comprends pas and je (ne) sais pas, which we regard as stereotyped non-segmented structures (cf. below in the text and Giuliano, in press, for a detailed discussion). . The expression isanot is employed by a learner of English in Huebner (1983). . For a detailed analysis of the negator [nepa(de)] in Bernarda, cf. Giuliano (in press). . PRO (pronoun) refers to the impersonal pronoun [se] of Spanish, that English translates by you/they. . Here our translation just tries to suggest the learner’s searching for the right verb inflection, since it is impossible for the analyst literally to translate the French realizations [kono / konos / konose]. . For a detailed analysis of the discursive functions of Italian prophrases, cf. Bernini (1995). . As for models, their attraction to INFL and the postmodal negation resulting from that are due to the nature of these verbs: modality is part of INFL (alias finiteness in our functional terminology). . Italian, differently from English, translates both sentences in (42) by the same sentence, Non ho soldi, (and so does Spanish) and does not permit utterances as *non idea (cf. angl. no idea and fr. aucune idée). . Cf. the Italian translation, nessuno dice mai niente per nessuna ragione. Both English and Italian exemples come from Bernini et Ramat (1992: 207). . Note that the overgeneralisation of no could be strengthened by transfer (cf. it. no vs. non), even though in contrast with the presence of complex negative forms in English (don’t, doesn’t, shan’t...). . In Table 3 below, we shall refer to this variety by the acronym PBV1. . Learners sometimes interpret irregular past tense forms in the input as present tense forms.



 Patrizia Giuliano . In Table 3 below, we shall refer to this variety by the acronym PBV2. . The symbol X refers variably to a noun, adjective or adverb. . This strategy is used by many L2 learners, who frequently create antonyms by the adjunction of a negative particle to a positive word. . The oblique form for me is an expression frequently exploited by Santo (cf. also for Sandra and for you). Klein (1981) remarks, for ZISA data, that in the least advanced interlanguages, bei mir refers to the speaker (“as to me”) rather than to a locative expression. . The numbers 1 and 2 refer to the intervention of two different interviewers. . This refers to a permit replacing the driving licence. . See the Conclusions for a further discussion of this point. . 11 stands for 1st cycle 1st encounter, 12 for 1st cycle 2nd encounter and so on. In other words, the first number always refers to the cycle whereas the second to the encounter. For the abbreviations in the columns, cf. Symbols and terminology at the end of the paper. . The arrow along the column refers to the progressive intensification of relational predicates. . This column contains the very first occurrences of finite verbs, already reported in Table 5. We must bear in mind that the emergence of some very rare instances of être/avoir and be/have at the basic stage has nothing contradictory if we consider the extremely dynamic character of any interlanguage, that slowly progresses from one stage to another. . Cf. Note 27. . For learners of English, don’t is employed as a non-analysed preverbal negator: Andrea will never master the suffix -(e)s of present tense and, in correlation with that, he will never master does not/n’t. Furthermore, although he employs the suffix -ed to express perfective or perfect actions in the past, he never produces negative utterances with didn’t. . Table 8 does not include the occurrences of is no possible, illustrated in Table 3 (cf. §5.4). . Cf. Note 27. . In Santo the relational predicates never become extensively productive (Aux + Vlex is limited to just one occurrence), which explains the basic character of his interlanguage along the whole data collection period. . This type of configuration is comparable to that described by Starren (2001), who has studied the use of Dutch is (“is”) and heeft (“has”) in the relatively advanced varieties of ESF learners of Dutch (Turkish and Arabic informants). These learners use these two verbal morphemes as aspectual markers inside VP, in particular, heeft for perfective and is for imperfective, as in the second occurrence of is in (i): Learner:

Nou is hij is bezem trekken Now is he is broom pull ‘He is pulling the broom’

These same markers can appear simultaneously with other inflected forms as the first is in (i) and was in (ii), referring to tense, which confirms L2 learners’ tendency to code complex

Negation and relational predicates

information (in this case, tense and aspect) by independent morphemes. The initial is/was occurs at the T/F boundary and marks the time for which the T/F relation is validated. Learner:

Die mesje was nooit heeft verkering The girl was never has relationship ‘The girl has never had any relationship’

. One might argue that if verb inflection initially emerges with predicates such as avoir/have and être/be, that happens for perceptual reasons and not for the more transparent functional nature of these verbs. Actually, in English and French, these verbs present a more morphologically varied paradigm with respect to lexical verbs. Yet, French irregular lexical verbs, even though having an equally varied paradigm, appear later with regard to avoir and être (for further details, cf. Giuliano, in press). Otherwise, many forms of be/have and être/avoir are not necessarily salient from a phonological viewpoint. In agreement with our results, Parodi (ZISA data) observes a precocious employment of the inflected forms of sein and haben but not of strong lexical verbs of German, and Bardel equally observes a precocious use of the inflection of Italian essere, in a language which also has many irregular verbs.

References Alarcos Llorac, E. (1994). Gramática de la Lengua Española. Madrid: Real Academia Española. Bardel, C. (2000). La Negazione nell’italiano degli svedesi. Sequenze acquisizionali e influssi translinguistici (Etudes Romanes de Lund 61). Lund, Sweden: The University of Lund. Bartning, I. (1997). L’apprenant dit avancé et son acquisition d’une LE. Acquisition et Interaction en Langue Etrangère, 9, 9–50. Becker, A. & Dietrich, R. (1997). The acquisition of scope in L2 German. Zeitschrift für Literaturwissenschaft und Linguistik, 26, 115–140. Benazzo, S. (2000). L’acquisition de particules de portée en Français, Anglais et Allemand L2. Etudes longitudinales comparées. Doctoral Dissertation, Université Paris VIII & Freie Universität Berlin. Bernini, G. (1992). Forme concorrenti di negazione in italiano. Linee di Tendenza dell’Italiano Contemporaneo, Les Actes du XXVème Congrès, Rome, Bulzoni. Bernini, G. (1995). Le profrasi. In L. Renzi, G. Salvi & T. Cardinaletti (Eds.), Grande Grammatica Italiana di Consultazione, Vol. III (pp. 175–222). Bologna: Il Mulino. Bernini, G. (1996). Stadi di sviluppo della sintassi e della morfologia della negazione in italiano L2. Linguistica et Filologia, 3, 7–33. Bernini, G. (2000). Negative items and negation strategies in non-native Italian. Studies in Second Language Acquisition, 22, 399–438. Bernini, G. & Ramat, P. (1992). La Frase Negativa nelle Lingue d’Europa. Bologna: Il Mulino. Cancino, H., Rosansky, E. J., & Schumann, J. (1978). The acquisition of English negatives and interrogatives by native Spanish speakers. In E. Hatch (Ed.), Second language acquisition. A book of readings (pp. 207–230). Rowley, MA: Newbury House.



 Patrizia Giuliano

Clahsen, H. (1988a). Critical phases of grammar development. A study of the acquisition of negation in children and adults. In P. Jordens & J. Lallemann (Eds.), Language Development (pp. 123–148). Dordrecht: Foris. Clahsen, H. (1988b). Parameterized grammatical theory and language acquisition: a study of the acquisition of verb placement and inflection by children and adults. In S. Flynn & W. O’Neil (Eds.), Linguistic Theory in Second Language Acquisition (pp. 47–75). Dordrecht: Reidel. Clahsen, H., Meisel, J., & Pienemann, M. (1983). Deutsch als Zweitsprache. Tübingen: Narr. Dahl, Ö. (1976). Typology of sentence negation. Linguistics, 17, 79–106. Dietrich, R. (1996). Neg-scope in a simple learner variety (Italian>German). Unpublished Manuscript. Nijmegen: Max-Planck-Institut für Psycholinguistik. Eubank, L. & Schwartz, B. D. (Eds.). (1996). Special Issue on The L2 Initial State. Second Language Research, 12(1). François, D. (1974). Français parlé. Paris: S.E.L.A.F. Gadet, F. (1992). Le français populaire. Paris: P.U.F. Giuliano, P. (1997). The acquisition of negation in French L2: a longitudinal study. In A. Sorace, C. Heycock, & R. Shillcock (Eds.), Proceedings of GALA ‘97 Conference on Language Acquisition (pp. 417–422). Edinburgh: The University of Edinburgh. Giuliano, P. (in press). L’acquisition de la Négation Linguistique dans une Langue Etrangère. (Un débat conclu?). Bern: Peter Lang. Harris, M. (1977). The evolution of French Syntax. A comparative approach. London: Longman. Horn, L. (1989). A Natural History of Negation. Chicago: University of Chicago Press. Huebner, Th. (1983). The Acquisition of English. Ann Arbor: Karoma. Kihlstedt, M. (1999). La granularité temporelle et le mouvement référentiel dans l’acquisition du français par des suédophones. Paper presented at the GRAL (Groupe de Recherche Acquisition des Langues) Meeting, Berder (France), 22–25 Mars. Klein, W. (1981). L’acquisition des pronoms personnels allemands par des travailleurs Espagnols et Italiens. GRECO, 13, 19–32. Klein, W. (1994). Time in Language. London: Routledge. Klein, W. & Perdue, C. (1992). Utterance Structure (Developing Grammars Again). Amsterdam: John Benjamins. Lardière, D. (1998). Dissociating syntax from morphology in a divergent L2 end-state grammar. Second Language Research, 14(4), 359–375. Manzotti, E. & Rigamonti, A. (1995). La negazione. In L. Renzi, G. Salvi, & A. Cardinaletti (Eds.), Grande Grammatica Italiana di Consultazione, Vol. III (pp. 245–320). Bologna: Il Mulino. Meisel, J. (1997a). L’acquisition de la négation en langue première. In C. Martinot (Ed.), Actes du Colloque International sur l’Acquisition de la Syntaxe en Lamgue Maternelle et en Langue Etrangère (pp. 189–222). Besançon: Analyses Littéraires de l’Université de la Franche-Comté. Meisel, J. (1997b). The acquistion of the syntax of negation in French and German. Contrasting first and second language development. Second Language Research, 13(3), 227–263.

Negation and relational predicates

Noyau, C., Houdaïfa, T., Vasseur, M.-T., & Véronique, D. (1995). The acquisition of French. In R. Dietrich, W. Klein, & C. Noyau (Eds.), The acquisition of temporality in a second language (pp. 145–209). Amsterdam: John Benjamins. Parodi, T. (2000). Finiteness and Verb placement in Second Language Acquisition. Second Language Research, 16(4), 355–381. Perdue, C. (1993). Adult Language Acquisition. Cross-linguistic perspectives. Vols. I & II. Cambridge: Cambridge University Press. Pollock, J. Y. (1989). Verb movement, Universal Grammar, and the Structure of IP. Linguistic Inquiry, 20, 365–424. Quirk, R., Greenbaum, S., Leech, G., & Svartvik, J. (1985). A Comprehensive Grammar of the English Language. London/New York: Longman. Schlyter, S. (1998a). Verbes et négation chez des apprenants suédophones du Français. Paper presented at the European Conference on Language Acquisition (European Science Foundation). Maratea (Italy), 28 September–1 October. Schlyter, S. (1998b). Négation et portée chez des apprenants suédophones du Français. Working Paper. Schlyter, S. (1999). Acquisition des formes verbales: temps, accord et finitude. Paper presented at the XI Colloque International sur l’Acquisition d’une Langue Etrangère: Perspectives et Recherches. Paris, 19–21 April. Schwartz, B. D. & Sprouse, R. (1996). L2 cognitive states and the Full Transfer/Full Access Model. Second Language Research, 12(1), 40–72. Starren, M. (2001). The Second Time. The Acquisition of Temporality in Dutch and French as a Second Language. Utrecht: LOT. Stauble, A. M. (1978). A comparison of a Spanish English and a Japanese English second language continuum: negation and verb morphology. In R. W. Andersen (Ed.), Second languages: A Cross-Linguistic Perspective (pp. 323–353). Rowley, MA: Newbury House. Stoffel, H. & Veronique, D. (1996). L’Acquisition de la négation en français par des adultes arabophones. Paper presented at the European Research Conference ‘The Structure of Learner Language. Utterance and discourse structure in language acquisition’, Espinho Portugal, September. Zanuttini, R. (1997). Negation and Clausal Structure. A Comparative Study of Romance Languages. New York/Oxford: Oxford University Press.



The copula in learner Italian Finiteness and verbal inflection Giuliano Bernini

.

Introduction1

This paper explores the position of the copula in the development of the verb system in second language acquisition of Italian. The major aim of the paper is to assess the function of constructions such as the one reported in (1) with respect to the expression of finiteness in learner language.2 (1) io [. . . ] sono è da razza cinese I am is from race Chinese ‘I am a Chinese’

(Peter, 9th rec.)3

Following the suggestions in Klein (1998, forthcoming), finiteness is considered here as a category distinct from, although related to, verbal inflection. In a finite utterance the descriptive content of the sentence is linked to a topic component and thus asserted as valid. The topic component includes the topic time, i.e. the time about which the assertion is made. Language acquisition, and in particular second language acquisition, may shed a light on the understanding of finiteness, which is almost inextricably fused with inflectional categories of the verb in many languages. In this paper it is argued that the use of the copula in non-target constructions as in (1) is one of the means by which learners start marking their utterances as finite. In this respect, the copula sono in (1) links the descriptive content of the sentence (è da razza cinese ‘being a Chinese’) to the speaker as a topic entity and to a topic time comprising the time of utterance, thus asserting that that content is valid in the real world. The paper is composed of five parts beside the introduction. In the first part (Section 2) the copula is dealt with in a typological perspective and the range of functions it covers across languages is discussed with particular respect to native and nonnative Italian and to Tigrinya, the Semitic language of Eritrea which is the

 Giuliano Bernini

first language of the learner to be considered in detail in this study. In the second part, the developmental path of the copula in this learner is described in order to identify potential stages in its acquisition.4 On the basis of the empirical investigation of the longitudinal data analysed in Section 3 and of previous literature, the major question relating to potential correlations between development of the copula and development of finiteness will then be approached in Section 4 and discussed with respect to negation in Section 5. In the concluding Section 6 the role of the copula in the acquisition of Italian L2 is assessed on the background of the development of verbal inflection and of utterance finiteness in post-basic varieties. Questions pertaining the position of the copula in language acquisition have been investigated within formal syntax with particular regard to German. Unlike in first language acquisition, in second language acquisition copula and modal auxiliaries as athematic verbs appear to develop finiteness and to favour the target postfinite verb position of negation before full thematic verbs, as claimed by Teresa Parodi (1998, 2000) in her study of bilingual children and adult acquirers of German.5 However, other researchers had already pointed at the role of the copula as a carrier of verbal categories in the acquisition of morphology, as e.g. Pfaff (1992) for the German L2 of Turkish children. As to Italian L2, use of the copula in nontarget analytic constructions was first described by Berretta (1990: 198–199), who interpreted them as alternative means for the expression of person and tense categories by learners who do not yet master verbal morphology. A strong correlation between acquisition of the copula and development of verbal inflection in Italian L2 has been more recently suggested by Bardel (2000), who pointed to the relevance of the copula for the acquisition of negation on the basis of the non-target post-finite verb position of negation in learners with Swedish as L1. In Sections 5 and 6 of this contribution previous suggestions on the copula in learner Italian will be assessed with respect to the notion of finiteness.

. The copula: Some typological considerations In a wider perspective, investigation of the copula in second languages is interesting for the simplification processes it may show with respect to baby talk, foreigner talk, pidgins and to full fledged languages too. In his seminal paper on the “Absence of copula and the notion of simplicity”, Ferguson (1971) established two major types of languages with regard to the parameter of copula expression. In Ferguson’s type A languages, such as English, Italian, and Tigrinya, positive equational clauses are marked by the copula, a particular type of verb which both functions as an auxiliary and appears in existential clauses of the type There is/there are. On the other hand, in Fergusons’s type B languages, such as Russian, Bengali and Hebrew, positive

The copula in learner Italian

equational clauses in the present are not marked, unlike non-present equational clauses and existential clauses, all showing different verb-like elements.6 The major constructions differentiating Ferguson’s type A and type B languages are shown in (2) for English and, respectively, Russian. (2) Type A: English a. The first boy on the left is Ivan

Type B: Russian

Pervyj first b. The first boy on the left is not Ivan Pervyj first c. The first boy on the left was Ivan Pervyj first d. There is a book on the table Est’ there.is e. There aren’t any books on the table Net there.is.not

mal’ˇcik boy mal’ˇcik boy mal’ˇcik boy knig-a book-nom.sg knig book(gen.pl)

sleva Ø on.left sleva on.left sleva on.left na on na on

Ivan Ivan ne Ø Ivan not Ivan byl Ivan was Ivan stol-e table-loc stol-e table-loc

In Russian, four different constructions correspond to the uses of the verb ‘be’ as the copula in English: in both positive and negative present equational clauses (2a) and (2b) no verb is used; in the past equational clauses (2c) a past form of the copula is used; finally, positive present existential clauses (2d) are marked by an existential verb and negative present existential clauses (2e) by the negative particle net, also used as a negative holophrastic reply. Different semantic functions are included under the heading “equational clauses” in Ferguson’s study (1971: 148, Note 1). Besides identity, as in example (1a), these functions are attribution of a property – e.g. The falafel is cold – and class membership – e.g. John is an expert. In the terms exposed by Émile Benveniste in his famous article on the linguistic functions of être and avoir (Benveniste 1960/1971: 225), the relevant question is the existence of a verb to be in type A languages, which give lexical expression to the abstract logical relation between the nominal terms of equational clauses unlike type B languages, which leave this unexpressed. This real “copulative” function of the verb to be must be distinguished from uses of to be in locative clauses of the type The book is on the table, where to be means to be situated, to lie, or of the type There is a book on the table, where to be means to have existence. Besides differing with respect to the parameter of the expression of the copula in equational clauses as discussed with regard to type A and type B languages in Ferguson’s terms, languages, and in particular type A languages, may also differ with respect to the range of locational functions for which the verb to be is used beyond the copulative ones – i.e. locative, existential and possessive, as inves-



 Giuliano Bernini

Table 1. Functional range of the copula in Tigrinya* and Italian (Present tense) Tigrinya

Functions

Italian

’6yyu ’6yyu ’6yyu ’6yyu ’allo ’allo ’allo

(Auxiliary) Identity Class membership Attribution of a property Locative Existence (Possession)

essere essere essere essere essere esser=ci avere

*Tigrinya data from (Kogan 1997: 444)

tigated by Clark (1978) on the basis of a sample of 40 languages and described in detail for a number of individual languages in the four volumes dedicated to “The verb ‘be’ and its synonyms” edited by John Verhaar (1967–1969). Different functional ranges of the copula may be illustrated on the basis of Tigrinya and of Italian. Tigrinya, a Semitic language of Eritrea, is the first language of Markos, the learner of Italian to be observed with regard to the acquisition of the copula; Italian is, of course, the target language to be investigated here. As shown in Table 1, Tigrinya has a major opposition between the verb ’6yyu, the copula used to mark identity, class membership, attribution of a property, and the verb ’allo, used in the locative and existential functions.7 On the other hand, Italian uses one single verb, essere for all the relevant functions listed in Table 1, although in the existential function it must be accompanied by the locative clitic ci, corresponding to English there. The difference in functional range of the copulas of Tigrinya and Italian may be illustrated on the basis of the examples reported in (3) for, respectively, the functions of class membership, attribution of a property, locative and existence. (3) Tigrinya* Italian a. Zar s6m färäs ’6yyu Zar è il Zar name horse is Zar is the ‘Zar is the name of a horse’ b. h.aww6-y d6ka ’6yyu Mio fratello brother-my poor is My brother ‘My brother is poor’ c. ’ab gäza-y ’alläw-u Sono a in house-my there-are-3pl.m they-are at ‘They are in my house’

nome di un cavallo name of a horse è povero is poor casa mia house my

The copula in learner Italian 

d. ’ab ‘addi .subbux. säb ’allo Nel villaggio c’ è un bravo uomo in village good man there-is in-the village there is a good man ‘In the village there is a good man’ *Tigrinya data (a), (b), (c) from Kogan (1997: 444); (d) from Conti Rossini (1940: 64) The Tigrinya copula ’6yyu shares with Italian essere the possibility of being used as an auxiliary: in the formation of the future in Tigrinya, in the formation of the perfect and pluperfect of a restricted set of verbs in Italian. On the other hand, the Tigrinya substantive verb ’allo is used for the expression of possession, a function reserved in Italian to a different verb avere, i.e. ‘to have’, as in other European languages. As shown in Table 2, use of essere for the functions listed in Table 1 is found in Italian in all tenses beside the present tense illustrated in (3) and in both positive and negative clauses. On the other hand, Tigrinya neutralises the opposition between ’6yyu and ’allo in the past and the future tenses: the past is formed in a suppletive way by means of the verb näbärä and the future is also formed in a suppletive way by means of konä. This same verb konä is also used for the negative counterpart of the copula, whereas the locative-existential predicate is negated by means of a separate verb yällon. After having established the major differences characterizing type A languages Tigrinya and Italian in the area of the copula, and the regular pattern of distribution of Italian essere across semantic functions, tenses and positive and negative clauses, it may be worth looking at the contact varieties of Italian as the target language to be considered, in particular at Italian foreigner talk and at Simplified Italian of Ethiopia as two contact varieties which may have an influence on the acquisition of Italian by Eritreans, in the host country and, respectively, in the home country. As a major feature of simplification expected to occur in foreigner talk varieties of type A languages according to Ferguson (1971: 146),8 in the elicited data of Italian foreigner talk investigated by Berruto (1993) the copula is deleted in just under

Table 2. Copula and locative-existential verbs: Negation and tense formation Positive Past Present Future

näbärä ’6yyu, ’allo konä

Tigrinya* Negative ay-näbärä-n ay-konä-n, yällon ay-konä-n

*Tigrinya data from Kogan (1997: 442, 444)

Positive (c’)era (c’)è (ci) sarà

Italian Negative non (c’)era non (c’)è non (ci) sarà

 Giuliano Bernini

fifty per cent of the cases and is the most frequent item to be restructured. A simple example is reported in (4). (4) Lei Ø tua sorella? she your sister ‘Is she your sister?’

(Berruto 1993)

As for the pidgin-like variety called Simplified Italian of Ethiopia, described by Habte-Mariam (1976), three items beside zero overlap in the expression of the functions listed in Table 2. These items, exemplified in (5), are: stare ‘stay’ for locative and attribution of a property; zero for identity; ce as a frozen form derived from the third person singular of the present indicative of the existential esserci for locative again and possession.9 (5) a.

loro stare addis abeba they stay-inf Addis Abeba ‘They live in Addis Abeba’ b. adesso loro stare amico now they stay-inf friend ‘Now they are friends’ c. kwesto Ø caße di borta? this key of door ‘(Is) this (the) key of (the) door?’ d. loro non ce kaza they not there.is home ‘They are not home’ e. iyo non ce/ aßere makkina I not there.is have-inf car ‘I do not have a car’

As a further remark on contact varieties, it is interesting to notice the use of stare in all functions but existence and possession, for which tenir ‘to hold’ is used, in the Lingua Franca as recorded in a dictionary published in Algiers in 1830 and edited by Cifoletti (1989). In learner Italian, and in particular in the Italian of the learner we are going to observe, stare does not belong to the set of forms used as a copula. Native input in the host country and not the Italian spoken in Eritrea appears therefore to guide the acquisition process.

The copula in learner Italian 

. The development of the copula in Markos For a first assessment of the potential paths of development of the copula in learner Italian, the 20 year old Eritrean learner called Markos was chosen. This learner was recorded for a period of seven months between October 1986 and June 1987, starting just after a month of stay in Italy. Recorded data comprise free conversations with an interviewer, narratives, and the retelling of a short movie called “The wallet”, specially produced for the collection of data on temporality within the “Pavia Project” on second language acquisition of Italian.10 Markos is mostly exposed to a varied and frequent input outside the public education program he attends and his development of Italian proceeds quite fast in the seven months during which he was recorded. The features of the Basic Variety are apparent in the first interview: the learner uses an invariable verb form which corresponds to the infinitive or to the third person singular of the present indicative, as vuole ‘wants’ in example (6); furthermore in the organisation of the utterance first position is reserved for the nominal argument whose referent is higher in control, as the government of Ethiopia in example (6) again, which is the source of the forced enlistment of the learner. (6) il governo de Tiopia- vuole io + militari the government of Ethiopia wants I military.man ‘I have to serve in the army under the Ethiopian government’ (Markos, 1 month, 1st recording) Features of a post-basic variety appear within the second month of stay in Italy since the third recording, with the establishing of the opposition of the unmarked verb form found in the Basic Variety and of a form corresponding to the past participle of the target with a past perfective meaning. In Markos’s first recording no copula is used in 15 contexts where the target language would require a form of the verb to be in all of the functions discussed before with the exception of existence as shown in Table 3, i.e. identity, class membership, attribution of a property, locative. For the functions of existence and possession c’è is used in 30 cases as an unanalyzed form corresponding to there is in the target, but functioning as an individual lexical item in learner language.11 The first recording includes two residual uses of c’è as a copula. In the first one, illustrated in (7), c’è is used as a locative predicate. (7) l’ uomo adesso c’è Amereca the man now there.is America ‘The man is now in America’

(Markos, 1 month, 1st recording)

 Giuliano Bernini

Table 3. Functions of the copula in Markos, 1st recording (1 month) Functions

Expression Examples

Identity

Ø

Class membership

Ø

Attribution of a property

Ø

Locative

Ø

Existence

c’è

(Possession)

c’è

questo Ø mio figlio ‘This is my son’ this my son lui-Ø Eritrea ‘He is an Eritrean’ he Eritrea Milano Ø bello sì’ ‘Milan is nice’ Milan nice yes lui adesso Ø America ‘He is now in America’ he now America non c’è lavoro ‘There is no work’ not there.is work (I: hai fratelli e sorelle?) non c’è you-have brothers and sisters not there.is ‘(Do you have brothers and sisters?) I do not’

In the second one, illustrated in (8), c’è is used as a copula in a rather peculiar contrastive context which may be classified as an instance of class membership. In fact, in this example, the learner corrects the Amharic word for a kind of cereal grass the interviewer has employed. (8) I:

così sì/ cioè col *t’ëf* si fa il pane so yes that-is with-the teff impers makes the bread ‘So, I mean, bread is made with teff ’ M: no non *t’ëf*- *t’ëf*- c’è *fernay* not teff teff there.is fernay ‘No. (It is) not teff, it is fernay’ [teff ‘Eragrostis abyssinica’] (Markos, 1 month, 1st recording)

Zero and the unanalyzed form c’è appear in positive and in negative clauses as well. In negative clauses the negator non is inserted between the nominal elements involved in the copular clause, as in (9), and in front of the existential c’è, as already shown in the last example displayed in Table 3. (9) I:

e poi- cioè Kasala è in- è ancora in Eritrea o è già and then that.is Kasala is in is still in Eritrea o is already in Sudan? in Sudan ‘And then, is Kasala in Eritrea or is it in Sudan’ M: Kasala?

The copula in learner Italian 

Kasala ah +++ eh +++ non- Eritrea- non Sudan Kasala ah eh not Eritrea not Sudan ‘Kasala? Kasala is not in Eritrea and is not in Sudan’ (Markos, 1 month, 1st recording) Absence of the copula irrespective of its presence either in the source or in the target language or both is a salient feature of the Basic Variety.12 In Ferguson’s terms (1971), Markos’s Basic Variety may be classified as a type B (learner) language, lacking a copula but possessing a means for the expression of existence, which is re-employed for possession too. The uses of the existential-possessive c’è in locative and class membership functions are too scanty and do not allow any consistent conclusion, except, perhaps the observation that use of c’è in the locative function recalls the functional range of the item used as a predicate of existence in the source language. The copula can be considered as acquired after eight months of stay in Italy. In the twelfth recording, the copula occurs in twenty eight over thirty possible contexts, and no copula occurs – as a remnant of former stages – in only two residual cases. Furthermore the correct morphological form occurs in each case as person agreement and tense are concerned and the functional range of the learner’s copula matches that of native speakers, including the auxiliary function in the formation of the compound past of a restricted set of verbs (cf. Table 1 above). On the other hand, essere is employed in connection with the locative clitic ci as a predicate of existence and the verb avere is employed for possession as in the target. An example for this final stage in the development of the copula is reported in (10). (10) (Commenting on regional varieties of Italian) M: perché di Napoli:, di: because of Naples of I: ah mh mh M: sono divers-i they-are different-pl ‘because (those from) Naples are different’ [= speak differently] (Markos, 8 months, 12th recording) In this final stage of the acquisition the learner also develops metalinguistic awareness of the orthographic peculiarity of the Italian copula in the present tense, as illustrated in (11). (11) l’ Italia è- verbo- con l’ accento the Italy is verb with the accent ‘Italy is (= in Italian this is the) verb with an accent’ (Markos, 8 months, 12th recording)

 Giuliano Bernini

The first occurrences of the copula are attested as soon as in the second recording, after one month and nine days of stay in Italy. The form found in this recording is recognizable as the third person singular of the present indicative è, which is pronounced with mid-low front vowel in the target. In this recording, the learner doesn’t seem to differentiate between mid-low and mid-high front vowels, the opposition found in the target language being neutralized in a mid front vowel in this learner’s language. This makes it difficult to distinguish copula forms from pause fillers and marginally from the coordinating conjunction e, which is pronounced with mid high front vowel in the target. Example (12) illustrates a case in point. In this example the items to be distinguished are reported in IPA transcription. All contain a mid front vowel. (12) questa famiglia ++ [’eœ:] ++ (li) figli [eœ_stu’deœnti] this family EH the children COP_students ‘This family eh the children are students’ [’eœ:] + padre de(l) figli + [’eœ:] + lavorare con, un ufficio E/EH father of-the children EH work-inf with a office ‘And/eh the father of the children eh works in an office’ (Markos, 1 month 9 days, 2nd recording) The guiding criteria in identifying pause fillers have been: (i) prosodic independence in terms of stress, (ii) presence of shorter or longer pauses around the item to be classified, (iii) potential length of the vowel, (iv) reduced volume as usual with parentheticals. Stress, pauses, length of vowel characterize the first item in IPA notation on line 1 and both items in IPA notation on line 4 in example (12). The last item on line 4 is also characterized as a parenthetical by reduced volume. As to the item in IPA notation in first position in utterance on line 4 in example (12), no criterion seems to help us classify it as a coordinating conjunction rather than as a filler. In contrast, items such as the second item in IPA notation on line 1 in example (12) lack prosodic independence, which implies that they are not separated by a pause from the following element, and they are thus better candidates for the role of the copula. The following example (13) shows the application of these criteria in the case of contiguous mid front vowels: the first one is a filler, the second one a copula. (13) *Zarai Dërës*, anche lui + [’eœ:] + [eœ_’grande]- [eœ_’grande] storìco, Zarai Dërës also he EH COP_great COP_great historic uomo man ‘Zarai Deres is a great man in history too’ (Markos, 2 months, 4th recording)

The copula in learner Italian 

Table 4. Expression of the copula in Markos Recordings

Length of stay (months, days)

Ø

essere

esserci

Overextension of essere onto esserci

1st 2nd 3rd 4th 5th

1, 0 1, 9 1, 22 2, 6 2, 22

15 4 13 9 20

– 3 5 20 20

2 4 2 – 14

– – – – –

6th

3, 12

7

52

–

9

7th 8th 9th 10th 11th 12th

3, 19 4, 12 5, 8 5, 29 6, 27 7, 28

– 5 – – – 2

15 28 25 9 20 29

– 1 1 2 1 –

2 2 4 2 2 –

In the subsequent recordings, vowel height is better distinguished by the learner and items such as the last one on line 1 of example (12) are better recognizable as copulas. After examining in some detail the methodological problems presented by the peculiar form of the Italian copula in the present tense, we can now consider the gradual development of the copula in Markos, on the basis of three sets of data. We shall consider first the evolution of zero copula along the recordings and move then on to the consideration of the copula forms attested in the learner’s corpus and included in the fourth column of Table 4; finally, the third kind of data to be considered comprise peculiar constructions. These will be analyzed in Section 3. As shown in the third column of Table 4, zero copula is found mostly in the first six recordings, till 3 months and 12 days since arrival in Italy. Essere as the main copula is established after two months of stay in Italy in the fourth recording; besides zero, its role is challenged by the existential c’è in the fifth recording at two months and 22 days, the existential being used mainly, but not exclusively, in locative contexts as the one exemplified in (7). However in the successive recordings the target form essere clearly prevails. The major turning point in the acquisition of the copula is mirrored in the sixth recording, when zero copula drops to seven occurrences against 52 occurrences of explicit copula. The development of the learner’s copula essere toward the target is also manifested in its employment as an auxiliary for perfect formation. Essere competes at first with avere in the role of the right auxiliary – 15 occurrences of essere against 9 occurrences of avere in the sixth recording – and with zero auxiliary, but appears to spread gradually onto the target contexts in the successive recordings.

 Giuliano Bernini

Overextension of essere to existential contexts beginning from the sixth recording, as shown in the last column of Table 4, and as exemplified in (14), appears to be connected with the morphological development of the paradigm of esserci rather than with the adjustment of their respective functional domains. (14) ieri anche + era congressa così yesterday also was convention so ‘Yesterday there was a convention too’ (Markos, 3 months, 12 days, 6th recording) The set of forms of the paradigm of essere attested in the learner’s corpus is displayed in Table 5. A restricted set of forms is developed by the learner in the period of observation, mostly belonging to the indicative present and imperfect – that is imperfective past. The three instances of the infinitive attested in the last recordings appear in connection with deve ‘must’, in epistemic reading. The most frequent forms in the learner’s corpus are the third person singular of the present and the third person singular of the imperfect. Present forms emerge after zero copula, as discussed before, and imperfect forms follow present forms, being attested from the fourth recording after two months of stay in Italy. On the background of the development of the copula in Markos as discussed so far, we can now turn our attention to the role the copula appears to have with Table 5. Development of essere in Markos Rec.

Rec. 1st 2nd 3rd 4th 5th 6th 7th 8th 9th 10th 11th 12th Tot.

Present Correct/overextended occurrences sono 1sg

sei 2sg

2/0

1/0

1/0

2/0

1/0 1/0 7/0

1/0

è 3sg 0/3 4/0 15/4 11/2 18/0 10/0 15/0 17/0 6/0 4/0 22/0 122/6

siamo 1pl

2/0 1/0

5/0 8/0

Imperfect Correct/overextended occurrences

sono 3pl

ero 1sg

1/0

1/0 5/0 4/0

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

0/1 1/0

9/0

11/1

Infinitive

era 3sg

erano 3pl

essere

15/4 5/0 10/0 4/0 2/0 5/0

3/4

2/0 1/0

2/0

41/4

8/4

3/0

2/0 1/0

N.b.: overextensions to esserci and auxiliary uses in the compound past are not included.

The copula in learner Italian

respect to the overall development of verb morphology in Italian L2, which can be summarised in the following two points: in the first place, the copula is the first, and for many learners the only, type of verb used in the imperfective past; in the second place, the copula is autonomously selected by some learners, including Markos, as a means for enriching the morphological potential of their postbasic stages. We consider in this section the first point and try to project the development of the copula onto the sequence of acquisition of tenses and moods. We leave to the next section the second point and consider there the peculiar uses of the copula illustrated at the beginning of this contribution. It will be recalled that the acquisition of the verb system of Italian involves the sequence of tenses and moods set out in (15).13 (15) Present/Infinitive > (Aux +) Past Participle > Imperfect > Future > Conditional [. . . ] However, whereas full verbs (especially non-durative) soon develop the past participle for the expression of the perfective past, they develop the imperfect much later (durative verbs first). In contrast, the copula develops the imperfect much earlier than other verbs. The relative timing of the emergence of copula and full verbs as attested in Markos is shown in Table 6. The copula enters Markos’s interlanguage at the early post-basic stage, when the first morphological opposition emerges, contrasting the uninflected forms derived either from the third present singular or from the infinitive of the target as the unmarked form and the past participle of the target marked for perfective (and in most cases for past too). Imperfect forms of the copula emerge after two months of stay in Italy, whereas first imperfect forms of full verbs are attested not earlier than after six months of stay in Italy.14 The leading role played by the copula in the emergence of the Italian Imperfect is to be related to the role that the Aktionsart (here, stative Aktionsart) of the verb Table 6. Sequence of tenses in Markos 1st occur. 1 month

1 month 9 days 3 months 12 days

Full verbs Present/Infinit. Past participle parla/studiare andato ‘speaks’/‘study’ ‘gone’

(Future) (no autonomous use of two lexical types, induced by school)

Copula

ero, era, etc. ‘I/(s)he was’ Imperfect

Ø

1st occur. 1month

è, etc. ‘is’ Present

1 month 9 days 2 months

6 months Imperfect sapevo, andavano ‘I knew’, ‘they were going’



 Giuliano Bernini

Table 7. Functions of essere in Markos Copula >

Aux (Present) in the compound past>

Aux (Present and Imperfect)

2 months è grande storìco uomo ‘(he) is a great man in history’

3 months, 12 days siamo parlato ‘(we) are [target: have] spoken’

3 months, 19 days erano andare ‘they were go [= going]’

plays in triggering the acquisition process of tense and aspect, i.e. the imperfective past in the present case. As suggested by Andersen (1991) for Spanish and shown by Giacalone Ramat (2002) for Italian, the Romance imperfect is acquired first with state and activity verbs, whose Aktionsart shares the feature of durativity with the aspectual value of the Imperfect.15 In Italian the present of the copula also has an auxiliary function in the formation of the compound past of a restricted set of unaccusative intransitive verbs, whereas with other intransitives and with transitive verbs the auxiliary avere – i.e. the verb of possession – is used. The two auxiliaries emerge simultaneously in the sixth recording (3 months and 12 days of stay in Italy), and early forms of past participles with no auxiliary – cf. Table 6 – are soon abandoned. In the process of diffusion of the two auxiliaries, essere is overextended in the formation of the compound past of verbs requiring avere more often than the reverse.16 We may then resume the development of the functions of essere in Markos’s interlanguage as illustrated in Table 7. Essere enters Markos’s interlanguage in the equational function and subsequently develops the target auxiliary function, accompanying the past participle in the formation of the compound past. When imperfect forms appear, essere is also used in a different non-target auxiliary function with verb forms other than the past participle, which will be the main concern of the following sections.

. Copula and verb morphology . Markos’s longitudinal data Besides being the first verb to emerge in the imperfect, the copula appears to be also relevant for the morphological development of Markos’s interlanguage, as shown by the peculiar non-target auxiliary constructions mentioned in the previous section, found between 3 months and 12 days and almost 7 months of stay in Italy, and mostly involving the imperfect of essere.17 We concentrate now on this type of data. Leaving aside the use of essere as an auxiliary followed by the past participle in contexts of perfective past (as in the

The copula in learner Italian 

target), we turn our attention to the use of: (a) essere with other verb forms (not past participle), (b) essere with past participle, but not in this strict sequence, i.e. where it does not immediately precede it (unlike in the target). These data include 13 occurrences of periphrastic formations of verbs, 10 involving the imperfect of essere and 3 the present. All are attested from the 6th recording onward. In the majority of these formations (11 out of 13) the copula precedes the lexical verb; in two occurrences the copula follows the lexical verb. In these last two occurrences the imperfect of essere follows c’è – one word for the learner – as the lexical existential verb in contexts where the imperfect of the existential is expected in the target. This strategy results in a two-words form c’è erano with stress both on c’è and on the first syllable of the second element, which nicely approximates the target form c’erano, where the original clitic ci fuses with the following form resulting in one phonological word with stress on the first syllable. This is illustrated in (16). (16) M: eh: non c’è, érano gli strumenti di musica *guitar* eh not there-is were the instruments of music ‘guitar’ ‘there weren’t any musical instruments’ I: ah non c’ érano ah not there were ‘there weren’t’ (Markos, 3 months, 12 days; 6th recording) The other occurrences to be considered here include formations where the copula is directly followed by an infinitive (4 instances), a gerund (1 instance), a present (2 instances), as illustrated in following examples. Notice in particular, in example (17), the opposition which seems to hold for Markos between the non-target periphrastic form era trovare, liter. ‘(he) was find-INF’ in a conditional context and the target periphrastic form ha trovato, liter. ‘he has found’ for perfective past. (17) eh: se- i documenti- non c’è erano di lui eh if the documents not there.is they were of him ‘if the documents weren’t his’ non/ non era: trovare tu/ telefono numero not not was find-inf telephone number ‘he wouldn’t find his telephone number’ ma però quando lui ha trovato il numèro but however when he has found the number ‘but when he found the number’ (Markos, 3 months, 19 days; 7th recording) (18) eh- lui con s/ con la sua moglie no? ‘he with his wife, right?’

 Giuliano Bernini

sta:/ erano andando al/ a giocare il tiennis stays(?) they-were going to-the to play-inf the tennis ‘they were going/went to play tennis’ (Markos, 6 months, 10h recording) (19) ha preso questo: portafoglio ‘he took this wallet’ allora ‘then’ era si chiama Giorgio no? was refl calls Giorgio no ‘he was called Giorgio’ [i.e. Giorgio is the name of the possessor of the wallet] (Markos, 6 months, 10th recording) In the remaining four cases the copula is followed by what looks like target periphrastic constructions. Three occurrences involve the progressive periphrasis with inflected stare followed by the gerund, as illustrated in (20). (20) I:

eh già, (xxx febbraio) hai cominciato ‘oh yes, February, you started’ M: sì ‘yes’ ancora st/ sono sto facendo still I-am I-stay doing ‘(and) I am still attending [this course]’ (Markos, 7 months, 11th recording)

In the last occurrence the present of essere precedes a negated compound past with have as the target auxiliary, cf. (21). (21) non non ha f:/ siamo non ha fatto not not has we-are not has done ‘We did not do (our show)’ (Markos, 3 months, 12 days; 6th recording) All examples (17) to (20) show the use of the copula as a means for the marking of some aspect of verb morphology. In particular, in example (17) the copula substitutes for imperfect morphology of an achievement verb, whose Aktionsart is not consistent with the durative feature of the imperfective past. However, the copula substitutes for imperfect morphology of other verb types, such as activities as in (18) and states as in (19) in the period where imperfect morphology of full verbs has yet to be developed, as shown in Table 6. On the other hand, the copula serves the aim of integrating new constructions into a morphological pattern, both imperfect and present, as in the case of the progressive construction with stare followed by the gerund shown in (20) (and perhaps (18) too).

The copula in learner Italian

Recalling the sequence of development of copula functions shown in Table 7, all constructions involving a form of essere may then be given a unified account. The copula is used in this learner’s language as an explicit link to finiteness with lexical elements which cannot incorporate finiteness as inflected verbs in the target language do. The copula is thus first used with nouns and adjectives (as in è grande storìco uomo ‘he is a great man in history’, cf. (13)) and with past participles as in the target language (cf. siamo parlato ‘we have spoken’). Subsequently, the copula is also used with any other verbal element which in the learner variety cannot yet incorporate the expression of finiteness (as in c’è erano, cf. (16), and in era trovare, cf. (17)). Tense distinctions in the copula (present è vs. imperfect era) code location of topic time within and before the time of utterance and, respectively, location of topic time after time of the situation with past participles and within the time of situation with other verb forms.18 These non-target uses of the copula may then be considered as instances of incipient grammaticalization of the copula in auxiliary function triggered by the way the copula may be used in the target language and can be dealt with in the framework of grammaticalization processes as discussed by Norbert Dittmar (1992) in general terms, by Giacalone Ramat (1992) for Italian and, with particular regard to the copula in the German of Turkish children, by Carol Pfaff (1992). Of particular interest in this respect is the last example mentioned above, where the copula precedes a negated compound past formed with the target auxiliary have. We delay the treatment of this occurrence until the next section, where negation will be at issue, after consideration of periphrastic formations recorded in the literature on Italian L2.

. Other learners Examples of autonomous formations involving the copula essere are attested in other learners, initial as well as advanced, beside Markos. Some of these examples have been already commented upon in different perspectives (e.g. Berretta 1990: 198–199; Berretta & Crotta 1991: 303–304; Bernini 1994: 285). One of these examples was already mentioned at (1). The role of the copula as the expression of a finiteness operator may be illustrated by the self-correction of the learner called Peter (cf. (22)), who rejects the simple form studiare – apparently infinite, i.e. non linkable to finiteness, in both the target and the learner language – in favor of a periphrastic form with the copula.19 (22) äh io studiare / I study-inf no no



 Giuliano Bernini

sono studiare a + I-am study-inf ‘I study’

(Peter, 1 month, 2nd recording)

Other categories beside tense/aspect and person for which the copula is used include conditional mood, as shown in the non-factual clauses reported in (23) and in (24) and drawn from the corpus of two advanced learners. In the first of these two examples, sarebbe as the target third person singular of the conditional mood of essere is adjoined to the imperfect of the same essere. In the second of these two examples, sarebbe appears in a complex construction resembling a cleft which includes the form avrà ‘(s)he will have’ inflected for third person singular future.20 (23) se era vero, sarebbe la persona era così if was true should-be the person was so ‘if this were true, that person should have been [positioned] so’ (Maria, ca. 9 months) (24) [on the contrary, if you work] non sarebbe che avrà dei problemi dopo not should-be that he-will-have of-the problems afterwards ‘you shouldn’t/won’t have problems later on’ (Alain, 2 years, 6 months) These constructions differ in one respect from the truly periphrastic ones found in Markos and in other learners. They both show initial position of the copula form, which either precedes the subject as in (23) or is clefted out of the sentence as in (24). However in both constructions the copula appears to be an independent expression for the non-validity of the propositional content of the sentence and therefore the role it plays in the coding of at least one of the components of finiteness is similar to the one shown for the periphrastic formations introduced above. A clue to the close link between the use of the copula as an independent finiteness operator and the development of verb morphology in Italian, where finiteness is carried by inflection, may be seen in the agglutination of the imperfect of essere to verbal stems in order to form the imperfect, attested in a couple of learners. Example (25) illustrates the use of era as a real copula in the first part of the utterance and, in the second part of the utterance, as a bound morpheme able to render the lexical verb vedere ‘see’ a finite form, resulting in the non-target form ved-ero ‘see-I was’ vs. standard ved-ev-o ‘see-Imperfect-1sg’. The interpretation of this form as a form inflected for the past is confirmed by the metalinguistic reflections of the learner.21 (25) in Dublino ero ehm senza lavoro e vedero, sì? ‘in Dublin I was without work and I saw, yes?’

The copula in learner Italian 

vedero? ‘I saw’ come si dice *I saw* vedero? ‘how do you say “I saw” vedero?’ (John, 2 months, 8 days; 3th recording) So far we have seen how the copula is used in learner Italian in order to construe finite utterances with lexical elements of different categories, nominal and verbal as well, in post-basic varieties. The sentence structures found in different learner varieties match the target structures in case of equational sentences and of past perfective sentences; otherwise they distinguish the learner variety from the target, where finiteness is coded by inflectional morphology on the verb. The use of the copula for the marking of finiteness is, as expected in learner language, variable both between different learners and within the interlanguage of inividuals. However, such non-target uses of the copula as those investigated here are of interest for the understanding of the development of finiteness and inflectional morphology in language acquisition. In particular, the relationship between non-target copula sentences and development of inflectional morphology is shown by the use of copula forms as bound morphemes on lexical verbs, found in some learners and illustrated above. One of the main aspect of finiteness in terms of Klein (forthcoming) is the reference to a topic time, i.e. the time span for which the assertion is valid. The position of the learners’ copula in this respect can be better established on the basis of a series of marginal examples involving negation and will be discussed in the next section.

. Negation The examples to be discussed in this section all involve non-target post-verbal position of the negator non with forms of the verb be, as in the example (21) recorded in Markos’s corpus, where the copula form siamo is followed by negation. The data at issue constitute a residue of exceptional cases in the acquisition of Italian negation, whose pre-verbal position doesn’t seem to be a difficult target, as discussed in Bernini (2000). These examples are scanty and marginal in the corpus of the “Pavia project” – cf. e.g. (26) – and more remarkable in the Italian of the Swedish adult learners investigated by Bardel (2000), where a verb fare ‘do’ in a copular use and the auxiliary have also trigger post-verbal negation beside the copula, as illustrated in (27).

 Giuliano Bernini

(26) nostra ca/ eh capodanno eh son eh no con italiano our New Year’s day they-are not with (= like) Italian ‘Our New Year’s day isn’t like in Italy’ (Chu, 1 year and 1 month ca.; 2nd recording)22 (27) a.

un problema de cambiatore è non buono ‘a problem of changer (?) is not good’ b. fare eeh hm no ingegnere ‘do [= be] not engineer’ c. ma eeh ho no fatto ‘but I-have not done’ (Bardel 2000: 112, 199, 114)

Four hypotheses have been suggested in order to account for these examples. In four cases of post-copular negation found in an advanced learner with German as L1 Bernini (1999: 114) has hypothesized that the negator non should not be considered as an instance of post-verbal negation, but should be related to the following adjective and thus be considered as an instance of analytic formation of antonyms, as not possible for impossible and not necessary for unnecessary reported in (28). (28) a.

perché + c’era non possibile andare: nelle strade ‘because there was not possible go-inf in the streets’ b. perché + ehm prima è non/ era non necessario parlare sui sol(di) ‘because before was not necessary speak-inf about money’ Antje (2 months)

Transfer was put forward as a second hypothesis by Bardel (2000) in order to account for the conspicuous presence of this kind of structure in learners with Germanic languages as L1s. This point of view is discussed by Bardel not only with respect to Swedish, the first language of the learners she investigated, but also to English as the major second language which might have influenced acquisition of Italian in her learners. According to Bardel, this type of transfer is observed when the learner variety already has some complex morphology. However, this type of transfer is limited to the copula and to the types of – athematic – lexical verbs mentioned above and in verb-complement structures; it never occurs with simple full verbs. In other words, Bardel’s Swedish learners never say *leggo non, calquing the corresponding Swedish structure Jag läser inte ‘I do not read’.23 According to a third hypothesis, suggested in Bernini (2000: 428) with regard to Chu’s example (26), post-copular negative structures attest an intermediate phase between an early pre-basic stage with negation before the focus constituent and the more advanced stage with pre-verbal negation found in basic and post-basic varieties. Finally, Bardel (2000) also suggested a correlation between the emergence of post-copular negative structures and the development of verbal inflection in the

The copula in learner Italian 

theoretical framework of formal syntax, whereby the position of the verbs before the negation mirrors the raising of inflected verbs to IP. On the basis of the evidence of the acquisition processes of languages with post-verbal negation as German, English and French, and following a suggestion put forward by Giuliano (2000), we may now try to suggest a kind of lectio difficilior of the data at issue. According to Giuliano (2000: 375), the different timing observed with athematic and with thematic verbs in the acquisition of post-verbal syntax of negation in French and in the Germanic languages may be explained by the fact that auxiliaries serve in a first phase to the coding of the topic time of the utterance. Since the span of the topic time is the one for which the propositional content of the negative utterance is valid, it is not included in the scope of negation and for this reason the form which marks it comes before the negative word. The constituent order of the resulting construction matches that of the target language, but obeys different principles. The Italian data at issue here may then be interpreted along the same lines as instances of the use of the copula as a marker of topic time, consistent with the data discussed in the previous section which pointed to the central role played by the copula in the establishing of finiteness when inflectional morphology is not yet acquired. Example (21) might be analysed as in (29) and paraphrased as an instance of perfect, with topic time after time of situation, as follows: “Concerning us, in the span of time talked about, which is located within the time of utterance, the situation of having-represented-our-show does not hold”. (29)

The uneven distribution of these examples among the learners, rare and isolated in general but for learners with a Germanic language as L1, can be accounted for in relation to native input on the one hand, where no examples of this kind occur, and to interference of the first language on the other hand, where post-verbal negative structures represent a widespread syntactic pattern. Transfer, however, is not the major principle involved and its manifestation is dependent upon processes of elaboration relating to the development of verbal inflection. As a result, different specific weights can be attributed to some of the factors mentioned in the hypotheses surveyed above. More importantly, the scanty number of utterances with non-target postcopular/post-auxiliary negation discussed in this section represent another piece of evidence in favour of the interpretation of the copula in initial – though not ex-

 Giuliano Bernini

clusively – post-basic varieties as one of the means resorted to for the marking of finiteness.24

. Concluding remarks The data discussed in the previous sections seem to point to a major relevance of the copula in the development of finiteness in terms of Wolfgang Klein’s suggestion (1998, forthcoming) and in the construction of the inflectional system of the target language, where finiteness is encoded. The copula seems to be chosen by the learners at different stages of acquisition as the carrier of the inflectional categories relevant for the establishment of finiteness when full verb morphology is not yet available, such as reference to time, reference to person, modality. Of particular interest in this respect is the employment of copula forms as bound morphemes on the one hand, and the evidence for the use of the copula as a marker of topic time delivered by its occurrence in non-target syntactic patterns with post-verbal negation. In general terms, these uses of the copula in negative utterances are the manifestation of processes in the acquisition of second languages which seems to be active irrespective of the syntax of negation of the target language – be it postverbal as German or pre-verbal as Italian. They throw a light on the role of the copula in the expression of finiteness as the carrier of the reference to topic time and help us in the identification of the different role played by different L2s (e.g. Italian vs. French and Germanic languages) in the dynamics of acquisition in the post-basic stage along routes diverging from the common ground characterising the Basic Variety. On this background let me make a last by-the-way remark about the paradox of finding in learner language widespread uses of the copula which seem to match the examples discussed in the history of philosophical thought and commented upon in the Appendix of Andrea Moro’s book on copular constructions (1997). Aristotle considered the copula an expression of tense and claimed that any sentence could be turned into a copular sentence (Moro 1997: 251). On the other hand Abelard, who actually introduced the term “copula” in his Dialectica in 1121, claimed that the copula is mainly a sign of affirmation, able “to turn a noun into a predicate”. The learner construction reported at the very beginning of this contribution nicely matches the sentence Socrates est ens, construed by Abelard in his speculations about the nature of the copula in the standard Latin sentence Socrates est.

The copula in learner Italian

Notes . The research presented here was financially supported by the Università degli Studi di Bergamo, grant “Fondo di ricerca d’ateneo”/2000, assigned to the Dipartimento di Linguistica e letterature comparate. I am indebted to Clive Perdue, Anna Giacalone Ramat, Pieter Muysken, Suzanne Schlyter, Christiane von Stutterheim and to the editors of the volume for their helpful comments and remarks. I am also indebted to Richard Dury (Università di Bergamo) for the revision of the English text. It goes without saying that any inconsistency is to be charged to the author’s stubborness. . Although dealing with comparable idiosyncratic constructions, this contribution and the contribution of Haberzettl in this volume look at them as the opposite faces of the same medal: as a first step towards the reconstruction of the target structures in the German of Turkish children studied by Haberzettl; as the autonomous elaboration of finiteness in the postbasic varieties investigated here. . Example recorded after 3 months and a half in Italy from a multilingual learner with Cantonese as L1 and Malay and English as L2. . The longitudinal data of this learner were collected within the “Pavia Project” on the acquisition of Italian. For a brief introduction to this project the reader is referred to Giacalone Ramat (1992). . Actually, the leading role of the copula in the acquisition of the syntax of negation in German L2 was first observed by Clahsen (1988). . In some languages this also holds for negative equational clauses. . As usual in the analysis and in the grammar of Semitic languages, the citation form of verbs used here is the third singular masculine form. . Cf. his “Hypothesis 1” which sounds: “In languages of type A, the copula in equational clauses will tend to be omitted in simplified speech such as baby talk and foreigner talk”. . In (5e) it is shown that ce competes with aßere ‘have’ in the expression of possession. According to Habte-Mariam (1976: 178), stare may be used for identity too, a function also expressed by e as a direct derivation of the Italian copula in the third person singular, but no examples are given. . The learner expatriated to Italy, where his mother had been working for more than ten years as a housemaid, in order to escape the war situation in the home country. After a couple of months in Italy he started working as an electrical contractor. During the period of observation, the learner attended a special public education program for foreigners; however his interlanguage does not show the normative features taught in schools apart from a little metalinguistic awareness. . Cf., in Table 3 non c’è lavoro and, respectively, non c’è used in response to a question of the interviewer containing the possessive verb avere. Convergence of expression of existence and possession is a widespread feature of second language acquisition, dealt with, among others, by Duff (1993). . In this respect, Markos’s initial learner variety of Italian is like initial varieties of the European languages investigated in the project on “Second Language Acquisition by Adult



 Giuliano Bernini

Immigrants” sponsored by the European Science Foundation (cf. Dietrich, Klein, & Noyau 1995: 262). . Cf. again Giacalone Ramat (1992) for a detailed consideration of this sequence. . Table 6 records early occurrence of two lexical types in the future in the sixth recording. These forms are triggered by school input; they are used with the interviewer by the learner, who wants to show what he has learned in the days preceding the recording, and are at once abandoned. As a consequence, use of these forms is not a counterexample to the general validity of the sequence illustrated in (15). As to the forms of the copula occurring in Markos’s interlanguage during the period of observation, the infinitive appears only later, in the tenth recording, as already pointed to in Table 5. . This question was also addressed in the Euroconference in the position paper on Temporality by Noyau (2000). . Actually, in the seven recordings at issue (6th to 12th), avere is overextended to essere in only one case, whereas essere is overextended to avere in 16 cases. In the sixth recording, e.g. following figures hold: ESSERE: zero (6), correct (15), overxtension to avere (9); AVERE: zero (7), correct (18), no overextension. . We leave aside 20 occurrences of the imperfect and 1 occurrence of the present of essere overextended in contexts where the existential esserci is expected. They have been already recorded in the last column of Table 4 and illustrated in (14). It may just be noticed in this regard that the past of the existential is left unexpressed until the fifth recording (at 2 months and 22 days of stay in Italy) and that generalisation of era ‘was’ in existential contexts starts in the sixth recording, i.e. at the same time as the target form c’era ‘there was’ emerges too. . The two forms of the copula in this learner’s language can be considered as FIN operators in the terms of Klein (forthcoming), applied to both nouns and verbs: FIN0 the present è; FIN< the imperfect era. For the interpretation of tense and aspect as resulting from combination of topic time with time of utterance and respectively of topic time with time of situation, cf. Klein (1994) and, in the perspective of second language acquisition, Dietrich, Klein and Noyau (1995: 22–25). . Peter is a 25 year old engineer from Malaysia with Cantonese as first language and Malay and English in his current repertoire (see also Note 3). He was recorded within the “Pavia project” for 8 months. Peter makes widespread use of periphrastic formations, cf. also siamo partenza (liter. we-are departure) ‘we left’; da Trieste siamo va in Varese, liter. ‘from Trieste we-are go-INF [= we go] to Varese’. A periphrastic formation involving the existential c’è is attested in the corpus of a 17 year old Australian student after 4 months and a half of stay in Italy: loro hanno telefonati io non ero c’è e Katia hai rispon/ hai rispondato? ‘They phoned; I wasn’t there and Katia answered’. . Example (23) was produced by an 18 year old Malawian nun with Chichewa as her first language; example (24) was produced by a 25 year old worker from the Ivory Coast with French and Moré as his first languages . John is a 27 year old Irishman with English as L1. Another imperfect form resulting from agglutination of era is sapera, produced by a 17 year old Australian student after 4 months and a half of stay in Italy.

The copula in learner Italian  . Chu is a 17 year old student, enrolled in a primary school in Turin. He speaks Wú Chinese as L1. . A remarkable real example of no transfer comprised in the data base of the “Pavia project” is the following: non + non le[ggo] le/ ++ *lese ich nicht* ‘I do not read [newspapers]’, produced by a 48 year old German learner after 1 year and 8 months of stay in Italy. . As to advanced varieties, an interesting example occurs as a hápax legómenon in the recording of an advanced learner with Egyptian Arabic as his first language and 10 years of stay in Italy, i.e. così non è mica c’ è Ghisa Cairo, (liter. ‘so not is not-at-all there is Giza Cairo’) ‘There is no Giza-Cairo’. In this example, which is worth noting even if only incidentally, the post-verbal negative particle mica, pragmatically marked and correct in the context of the conversation from which the example is taken, is not positioned directly after the main verb c’è, as expected in the target, but inserted into the utterance by means of the auxiliary copula è ‘is’. The learner seems to overcome the difficulty of using a postverbal negation by attaching it to an auxiliary copula, reflecting the regularities found in the acquisition of languages with postverbal negation.

References Andersen, R. W. (1991). Developmental sequences: the emergence of aspect marking in second language acquisition. In Th. Huebner & Ch. A. Ferguson (Eds.), Crosscurrents in Second Language Acquisition and Linguistic Theories (pp. 305–324). Amsterdam: Benjamins. Bardel, C. (2000). La negazione nell’italiano degli Svedesi. Sequenze acquisizionali e influssi translinguistici. Lund: Romanska Institutionen (Études romanes de Lund 61). Benveniste, É. (1960). ‘Être’ et ‘avoir’ dans leurs fonctions linguistiques. Bulletin de la Société de Linguistique, 55, 113–143. (Quotation according to the Italian edition: ‘Essere’ e ‘avere’ nelle loro funzioni linguistiche. In E. Benveniste, Probemi di linguistica generale (pp. 223–247). Milano: Il Saggiatore (1971)). Bernini, G. (1994). Le frasi ipotetiche nell’italiano di stranieri. In A. Giacalone Ramat & M. Vedovelli (a cura di), Italiano. Lingua seconda/lingua straniera (pp. 271–296). Roma: Bulzoni. Bernini, G. (1999). Percorsi di apprendimento della negazione. In N. Dittmar & A. Giacalone Ramat (a cura di), Grammatik und Diskurs / Grammatica e discorso. Studi sull’acquisizione dell’italiano e del tedesco / Studien zum Erwerb des Deutschen und des Italienischen (pp. 91–124). Tübingen: Stauffenberg. Bernini, G. (2000). Negative items and negation strategies in non-native Italian. Studies in Second Language Acquisition, 22, 399–438. Berretta, M. (1990). Morfologia in italiano lingua seconda. In E. Banfi & P. Cordin (a cura di), Storia dell’italiano e fomre dell’italianizzazione (pp. 181–201). Roma: Bulzoni. Berretta, M. & Crotta, G. (1991). Italiano L2 in un soggetto plurilingue (cantonese – malese – inglese): sviluppo della morfologia. Studi Italiani di Linguistica Teorica e Applicata, 20, 285–331.

 Giuliano Bernini

Berruto, G. (1993). Italiano in Europa oggi: «foreigner talk» nella Svizzera tedesca. Omaggio a Gianfranco Folena, 2275–2290. Padova: Editoriale Programma. Cifoletti, G. (1989). La lingua franca mediterranea. Padova: Unipress. Clahsen, H. (1988). Kritische Phasen der Grammatikentwicklung: Eine Untersuchung zum Negationserwerb bei Kindern und Erwachsenen. Zeitschrift für Sprachwissenschaft, 7, 3–31. Clark, E. V. (1978). Locationals: Existentials, Locative, and Possessive Constructions. In J. H. Greenberg et al. (Eds.), Universals of Human Language: Syntax, Vol. 4 (pp. 85–126). Stanford: Stanford University Press. Conti Rossini, C. (1940). Lingua tigrina. Parte prima. Elementi grammaticali ed esercizi. Roma: Mondadori. Dietrich, R., Klein, W., & Noyau, C. (1995). The acquisition of temporality in a second language. Amsterdam: Benjamins. Dittmar, N. (1992). Grammaticalization in second language acquisition. Introduction. Studies in Second Language Acquisition, 14, 249–25. Duff, P. A. (1993). Syntax, semantics and SLA: The convergence of possessive and existential constructions. Studies in Second Language Acquisition, 15, 1–34. Ferguson, Ch. A. (1971). Absence of copula and the notion of simplicity: a study of normal speech, baby talk, foreigner talk, and pidgins. In D. Hymes (Ed.), Pidginization and creolization of languages (pp. 141–150). Cambridge: Cambridge University Press. Giacalone Ramat, A. (1992). Grammaticalization processes in the area of temporal and modal relations. Studies in Second Language Acquisition, 14, 297–322. Giacalone Ramat, A. (2002). How do learners acquire the classical three categories of temporality? In Y. Shirai & R. Salaberry (Eds.), Description and Explanation in the L2-Acquisition of Tense-Aspect Morphology (pp. 221–248). Amsterdam: Benjamins. Giuliano, P. (2000). L’acquisition et l’expression des fonctions négatives en français et en anglais comme langues secondes. Confrontation d’études longitudinales et apports théoriques pour l’acquisition en milieu naturel. Thèse de doctorat en co-tutelle. Université de Paris VIII/Università degli Studi di Pavia. Habte-Mariam, M. (1976). Italian. In M. L. Bender et al. (Eds.), Language in Ethiopia (pp. 170–180). London: Oxford University Press. Klein, W. (1994). Time in language. London: Routledge. Klein, W. (1998). Assertion and finiteness. In N. Dittmar & Z. Penner (Eds.), Essays in Honor of J. Weissenborn (pp. 225–245). Bern: Peter Lang. Klein, W. (forthcoming). On finiteness. In V. van Geenhoven (Ed.), Semantics meets acquisition. Dordrecht: Kluwer. Kogan, L. E. (1997). Tigrinya. In R. Hetzron (Ed.), The Semitic Languages (pp. 424–445). London: Routledge. Moro, A. (1997). The raising of predicates: Predicative noun phrases and the theory of clause structure. Cambridge: Cambridge University Press. Noyau, C. (2000). Les relations temporelles dans la dynamique des lectes d’apprenants. Unpublished manuscript presented at the Euroconference The Structure of Learner Language. Information structure, linguistic structure and the dynamics of acquisition, San Felíu de Guíxols (Spain), 6–10 October 2000.

The copula in learner Italian

Parodi, T. (1998). Der Erwerb funktionaler Kategorien im Deutschen: Eine Untersuchung zum bilingualen Erstspracherwerb und zum Zeitspracherwerb. Tübingen: Narr. Parodi, T. (2000). Finiteness and verb placement in second language acquisition. Second Language Research, 16, 355–381. Pfaff, C. (1992). The issue of grammaticalization in early German second language. Studies in Second Language Acquisition, 14, 273–296. Verhaar, J. M. W. (1967–1969). The verb ‘to be’ and its synonyms, Parts 1–4. Reidel: Dordrecht.



The interaction between the development of verb morphology and the acquisition of temporal adverbs of contrast A longitudinal study in French, English and German L2* Sandra Benazzo

.

Introduction

Previous studies on the expression of temporality have highlighted the interplay of various devices expressing temporal relations and their relative weight at different stages of second language acquisition. Generally speaking, it has been observed that, as long as the verb is not functionally inflected, temporal relations are mainly expressed by discourse organisational principles and adverbials, which allow learners to quantify over events or to locate events on the time axis (Dietrich et al. 1995). It has even been hypothesised that a clever handling of adverbs could prevent learners from progressing further and developing grammatical means for tense and aspect marking (Sato 1990; Starren & Van Hout 1996). In spite of the important role assigned to temporal adverbs in early stages of acquisition, temporal adverbs of contrast such as again, already, still, yet (from now on TACs) seem to be mainly attested in the production of advanced learners (Dietrich et al. 1995: 266). This late acquisition has been explained on the basis of their conceptual complexity: TACs relate and contrast different time intervals of a state or an event (cf. next section), so that they can be considered to express cognitively complex temporal configurations. The statement about a generally late acquisition of TACs needs in fact to be qualified, as recent research indicates that some of them can appear, combined in a creative idiosyncratic way, at earlier stages (Starren 2001). Such observations suggest that the L2 acquisition of this adverbial category needs further investigation: it still is not clear at which stage they start appear-

 Sandra Benazzo

ing in learner production, how learners make use of them, what their place is among other devices to express temporal relations and indeed why, in spite of their adverbial status, they should typically belong to the repertoire of advanced learners. Moreover TACs can be considered as scopal items whose meaning changes, providing temporal (1a) or non-temporal readings (1b), as a function of the constituent in their scope: (1) a.

Il boit encore he still drinks, he drinks again b. Il boit encore un café he drinks another/one more coffee

Given their scope properties and/or their polysemy in the TL, it is interesting to observe how learners deal with their temporal meaning among other possible ones and what the link is between their acquisition and the acquisition of other scopal items. In order to pursue this subject, we have analysed the longitudinal data of learners of English, French and German L2, which belong to the ESF data-base (European Science Foundation’s Project on Second Language Acquisition by Adult Immigrants, Perdue 1993). The research presented here is part of a larger study on scopal items in L2 (Benazzo 2000), where we observed the acquisition of additive, restrictive and temporal particles. The present paper aims to analyse the emergence and use of TACs, in relation to other scopal items and in relation to the general development of the L2 system. In the following sections, before analysing TACs in L2 (sections 4–5), we will briefly define (Section 2) the temporal configuration they express and describe (Section 3) the linguistic development of the learners analysed during the observation period.

. Temporal adverbs of contrast and other scopal items Items such as still, already, again (and their negative equivalents not yet, anymore, no longer) form a special category of temporal adverbs (Klein 1994) which encode complex time configurations.1 In order to observe their semantic contribution, consider the following sentences: (2) a. at 10 John was still sleeping b. at 10 John was sleeping again c. at 10 John was already sleeping

Verb morphology and the acquisition of temporal adverbs of contrast 

In comparison to the explicit time interval “at 10” at which the state-of-affairs “John’s sleeping” obtains, the presence of still implies that the corresponding event (John’s sleeping) continues from a previous adjacent time interval (“continuative” reading); again means on the other hand that the same event has taken place in a previous but not adjacent time interval (“iterative” meaning), while already marks the transition from a negative to a positive phase of the same event (“resultative” reading):2 it implies that John wasn’t sleeping in a previous undetermined time interval. The basic semantic contribution of these adverbials can be defined as a quantification over alternative time spans for a given state or event. While in English each item is somewhat specialized for a certain semantic time configuration, in other languages, for example French, some items are multifunctional: this is the case of encore (cf. Borillo 1984; Franckel 1989) which corresponds to both still and again: (3) il dort encore he is still sleeping / he is sleeping again The distinction between an iterative or continuative reading is then mainly based on the aspect of the predicate: imperfective aspect implies continuative reading (il dormait encore ‘he was still sleeping’), while perfective aspect implies iterative reading (il a encore dormi ‘he has slept again’). In other contexts, the same item can also correspond to more: (4) il dormira ici encore une nuit he will sleep here one more/another night In (3) encore is a temporal adverb affecting the predicate, whereas in (4) it is used as an additive particle quantifying over the NP-constituent one night. Similar scope properties also characterize German noch/schon – which can be used with a temporal meaning (equivalent to ‘already/still’) or as additive scalar particles (equivalent to ‘even’) – and English just, which occurs with a restrictive scalar meaning (‘he ate just one apple’) or temporal meaning (‘he’s just gone’). Temporal as well as nontemporal (additive/restrictive) readings can be accounted for by the fact that TACs and additive/restrictive particles share the common property of quantifying over alternatives, although for TACs these alternatives are limited to time intervals (for more details cf. König 1977 and 1991; Löbner 1989; Muller 1975). Given the versatility of each item, it is necessary to look carefully at the discourse context where they occur in order to determine with which value they are being used.

 Sandra Benazzo

. Learners and their development The emergence and use of TACs is investigated in the language production of learners first studied in the European Science Foundation’s project: “Second Language Acquisition by Adult Immigrants” (Perdue 1993). The following longitudinal data have been analysed: 3 Italian learners of English (Santo, Andrea, Lavinia) 2 Italian learners of German (Tino, Angelina) 3 Spanish-speaking learners of French (Bernarda, Palmira, Alfonso) The informants are all adults who had settled in the TL-country for economic and/or political reasons and have been recorded monthly over a period of about 30 months. The recordings are organised in 3 data-collection cycles, each cycle corresponds to a series of communicative activities which were accomplished in a fixed order and repeated in the following two cycles (for more details, see Perdue 1993). At the beginning of the recordings most of them just knew a few words of the target language, that is they were ‘real’ beginners, while at the end of the observation period their proficiency in the target language is quite variable: as attested in previous studies (for example Klein & Perdue 1992; Dietrich et al. 1995), some of them reach a very advanced stage (Tino, Lavinia, Alfonso), others “fossilize” early (Angelina, Santo). In order to make crosslinguistic comparisons, in the following table they are classified according to the acquisitional progression proposed by Klein and Perdue (1997) in 3 main stages: prebasic, basic and postbasic. Table 1. Learners’ second language development

Some learners’ production initially oscillates between prebasic and basic variety (Santo, Andrea and Tino), while the attainment of a postbasic stage is limited for some learners to the development of selected features (Andrea, Bernarda).

Verb morphology and the acquisition of temporal adverbs of contrast

Each stage corresponds to a specific organisation of the learner’s system, concerning utterance structure as well as the expression of temporal relations. At the prebasic level, learner utterances mainly consist of nominal constituents (roughly corresponding to TL nouns, adjectives, adverbs), which are related on the basis of pragmatic principles; temporal reference can be reconstructed from the interlocutor’s contribution (scaffolding), inferred from discourse organisation principles such as the Principle of Natural Order (= events are related following their chronological order) or expressed by some adverbials, especially calendaric expressions. The following stage, the basic variety, is marked by an organisation of the utterance around a non-finite verb and its arguments. At this stage verb morphology is dysfunctional (no grammatical marking of tense or aspect), but the presence of verb-argument structure permits the expression of inherent lexical aspect (Aktionsart), expressions equivalent to start/finish signal event boundaries, while a large repertoire of temporal adverbs – of position (4 o’clock, yesterday, last year), of frequency (three times, once a week) and of duration (for one hour) – allow learners to locate events in time and to quantify their frequency or duration, as can be seen in the following passage. (5) monday i gone in the hospital and speak doctor for i chop my nail last month i going hospital two hours in the room operation

(Santo: cycle 2, rec. 22)

The postbasic variety is characterized by the emergence of finite verb morphology, which thus expresses grammatical tense and aspect. The transition from the appearance of the first functional morphological oppositions to their systematic application is usually a slow and gradual process: for the learners analysed here, finite verb forms first appear on the copula or the auxiliairies, before spreading to lexical verbs (Perdue et al. 2002). Given that at this stage learners’ paths differ as they approach the TL specifics, further details on the progression stated for each learner will be given in the next sections, during the analysis of TACs. The following tendencies concern all of them however (cf. Dietrich et al. 1995): – – –

initial coexistence of various morphological forms without appropriate functions; tense marking precedes aspect marking; irregular morphology precedes regular morphology.

The expression of temporal relations is then marked by a crosslinguistic development sequence where pragmatic devices (discourse organisation principles), pre-



 Sandra Benazzo

cede lexical ones (temporal adverbs, lexical content of verbs, etc.), which in their turn precede grammatical ones (verb inflection). The passage from one stage to the next provides new means for the specification of more fine-grained temporal relations. Starren (2001) has for example pointed out which temporal relations adverbials can or cannot express according to the degree of grammaticalisation of the verb. She found that the lexical devices applied at the basic variety level allow learners to express a dense web of temporal and aspectual relations, but events are invariably more or less simultaneous to the lexically specified time span. In order to dissociate them, as in prospective aspect (focus on the pre-time of the relevant event) or perfect aspect (focus on the post-time of the relevant event),3 learners need to develop verb morphology.

. Overview on the repertoire of the L2 scopal items Learners’ progression towards the TL is marked by a parallel use of an ever wider range of scopal items. The following Tables (2–4) give a preliminary overview of this process: for each learner the columns represent the spontaneously produced particles (direct repetitions of an item from native speaker input and uninterpretable utterances are excluded) and the horizontal lines, the sequence of recordings. As these are organised in 3 data-collection cycles, the first number indicates the cycle, the second number the recording (15 means 1st cyle, 5th recording). The contextual meaning of multifunctional items is identified by letters (A= additive, It = iterative, C = continuative), while idiosyncratic uses are marked by asterisks. There are correspondences between the acquisitional stage and the type of particles used by the learners, hence the acquisitional stage is given on the top line of the tables. We first comment on the Italian learners of English. At the beginning of the observation period their production shows the typical features of the basic variety: most utterances consist of non-finite verbs with their arguments and, optionally, adverbials, although some traces of a previous prebasic level are still present in Santo and Andrea. Santo does not go beyond this stage: even if the verb starts showing some morphological variation, there is no clear functional marking of tense or aspect before the last recording (as example (5) clearly shows). As can be seen in Table 2, he makes use of a limited range of particles: restrictive and additive ones (such as only, just, also, too), and one temporal item, that is again. In Santo’s production again principally marks the reiteration of a state or an event (6a), but sometimes it is used as an additive particle (6b) and probably also with other meanings (6c).

Verb morphology and the acquisition of temporal adverbs of contrast 

Table 2. Repertoire of scopal items: Learners of English

*Idiosyncratic uses: some occurrences of Santo’s again seems to have an additive value equivalent to also, more; Andrea’s again is produced in contexts of continuity, thus requiring still; Andrea’s still has not an adverbial status, functioning as an idiosyncratic auxiliary.

(6) a.

(free conversation: the NS has just come back from holiday) when going in holiday again?4 (Santo: cycle 1, rec. 17) b. (free conversation: the interlocutor has told Santo that he knows a few words of Italian) again (= more) 20 years and you speak Italian (Santo: cycle 1, rec. 19) c. (Santo is waiting for his driving licence) maybe i wait again another month = I keep on waiting? I wait one more month? Once again I wait one month? (Santo: cycle 1, rec. 16)

 Sandra Benazzo

At the beginning of the recordings, Lavinia’s and Andrea’s productions also roughly correspond to the basic variety and, as Santo, their initial repertoire of particles includes one productive temporal item, that is, again. But, unlike Santo, these learners progress further. During the 2nd and 3rd cycle, Andrea shows some signs of development towards a postbasic variety, although this is limited to the progressive development of a functional opposition between present and past verb forms. At this stage he first shows some idiosyncratic uses of again, which is overgeneralised to contexts requiring a continuative adverb like still, and later on some occurrences of still, which is however not used in the adverbial function (hence the question mark after it in Table 2). These idiosyncratic uses will be commented on later. Lavinia’s progression towards a postbasic variety ends up with her mastering of the English verb morphology. This evolution coincides with the production of many different temporal items: already, still, just (in the temporal function, see the 2 occurrences marked by ‘T’ in Table 2), not. . . yet and not. . . anymore (cf. (7a–e)). (7) a. I was just going out to the park when I see this (Lavinia: cycle 2, rec. 21) b. (the NS asks Lavinia whether she has taken her examination) I have to + no yet (Lavinia: cycle 2, rec. 22) c. my landlord already + did + this kind of thing (Lavinia: cycle 2, rec. 23) d. (the NS asks Lavinia whether Yugoslavs come over to Italy quite a lot) they used to, but not anymore (Lavinia: cycle 2, rec. 24) e. Carlo is still working (Lavinia: cycle 2, rec. 26) The comparison of the three Italian learners of English reveals a certain correlation between the type of particles used by leaners and their acquisitional stage: the iterative again is attested very early in the production of all of them, while other items appear (as TACs) only in the production of the more advanced learner. A similar order of acquisition is also attested for learners of French and German L2 (cf. Tables 3 and 4). The initially less advanced learners (Palmira and Bernarda for French L2 and Angelina for German L2) provide evidence from the very beginning of the acquisitional process: the first items to appear are namely restrictive (seulement, nur, allein) and/or additive particles (aussi, encore, auch, noch) which are produced from the pre-basic variety onwards, thus preceding the emergence of iterative items. These are sometimes idiosyncratic forms – as autre fois in French L2 instead of the TL encore/de nouveau5 – or idiosyncratic extensions of already available additive items – as auch in German L2 instead of TL nochmal/wieder – but anyway attested with an iterative function from the level of the basic variety onwards.

Verb morphology and the acquisition of temporal adverbs of contrast 

Table 3. Repertoire of scopal items: Learners of French

Learners who progress further (Bernarda, Alfonso, Tino) enrich this repertoire with items equivalent to ‘already’ and ‘still’: this means either the appearance of a new item (déjà/schon, ‘already’) or the specialization of the functions of an already available item (noch /encore) which was first used to mark the addition of entities (‘more’, cf. (8a)), and last of all for time intervals (‘still’, cf. (8b)). (8) a.

noch für ein Monat for one more month b. ich bin noch böse I’m still angry

(Tino: cycle 2, rec. 25) (Tino: cycle 2, rec. 27)

The negative counterparts equivalent to not yet, no longer, no more are, as in English L2, late marginal acquisitions, often limited to formulaic expressions and therefore not counted here.6

 Sandra Benazzo

Table 4. Repertoire of scopal items: Learners of German

*Idiosyncratic uses: some occurrences of Angelina’s auch and of Tino’s noch seem to have an iterative value equivalent to nochmal.

The comparison of the three groups of learnes highlights a fixed order of acquisition: after additive and restrictive items, learners start producing iterative items; only later on do they gradually develop TACs equivalent to (resultative) ‘already’ and (continuative) ‘still’. Leaving aside the variable frequency of these items, this order of appearance correlates crosslinguistically with the general development of the learners’ production, as it is summarised in Table 5. Precocious use of items classified here as belonging to the postbasic-variety level – namely continuative noch in Angelina (1 occ.) and déjà in Palmira (1 occ.) – do not run counter the general picture of this diagramme, as they are very rare and do appear after additive/restrictive and iterative items. It is however useful to notice their presence, which will be further commented in the next section. Table 5. Enlargement of the L2 scopal repertoire Additive Restrictive

>

Fr. seulement, aussi, encore Engl. only, just, too, as well Ger. auch, allein/nur, noch PREBASIC V.

Iterative autre fois again nochmal BASIC V.

>

Resultative Continuative déjà, encore already, still schon, noch POSTBASIC V.

Verb morphology and the acquisition of temporal adverbs of contrast 

. Learners’ use of TACs in discourse In order to explain the acquisitional sequence shown in the tables, it is necessary to look at the discourse functioning of these particles and at the devices used by learners to express temporal relations.

. Iterative items at the basic variety Iterative items appear at the basic variety level. At this stage temporal relations are mainly expressed by lexical means (temporal adverbials and the lexical content of the verb), otherwise they are situated by default at the time of utterance or inferrable on the basis of shared knowledge (the expression when in Italy, produced by an Italian learner, is interpreted with reference to the past such as when I was in Italy). At this stage learners use iterative items sometimes in association with a temporal adverbial (for example Andrea: six o’clock again go in Holborn), or with other expressions implying temporal reference, as in the following passage: (9) (Santo has paid a fine for having crossed a street while the trafic light was red) NS what you paid fifty pounds for? for traffic light .... is very expensive for traffic light possible twenty and next ++ next traffic light me again *non* (= don’t) see the traffic light possible fifty one hundred pound (Santo: cycle 1, rec. 18) In (9) the event ‘pay fifty pounds’ is first situated in the past (temporal anchoring maintained from the NS question) and then in a hypothetical future. Given that verb morphology is still dysfunctional, the time span which separates the two occurrences of the same situation is given by the unit ‘next traffic light’, which is roughly interpreted as ‘next time I cross a red traffic light’. But iterative items are more frequently present in narratives, as in the following extracts, and without any adverbial specification: (10) a.

(film retelling: reporting a scene where different people try to open a locked door) another man *prove* for er open the door but no possible *allora* that woman er/ this woman phone the *brigate* fire for help. . . another woman . . . *prove* again for open the door (Andrea: cycle 1, rec. 13)

 Sandra Benazzo

b. (film retelling Modern Times: the police arrests the girl who has stolen a bread) die polizei fliehe the police run nehme die mädchen mit brote take the girl with bread die polizei fliehe nochmal the police run again (Tino: cycle 1, rec. 17) c. (film retelling Modern Times: two times a beam falls onto Chaplin’s head) le tombe une table sur la tête him (C.C.) falls a beam on the head (series of actions) monsieur chaplin rentre à mister chaplin goes into la maison the house *y otra* fois le tombe le table sur and again him falls the beam on la tête the head (Palmira: cycle 3, rec. 34) The basic temporal meaning of again implies that the relevant event designed by the predicate has already taken place before, in a previous but not adjacent time interval. In all of these passages the antecedent event is mentioned in the preceding co-text and the iterative item creates an anaphoric link to it. In this way, from a discourse point of view, iterative items behave like additive particles, which establish a similar additive link, but prototypically between entities.7 From a temporal point of view the repetition of an event involves two tokens of an event of the same type, temporally ordered and not adjacent. In order to be repeated (instead of, for example, continued) an event has to be bounded. In highly structured texts such as narratives, this is achieved by simple adherence to the principle of natural order, which is regularly applied: respecting this pragmatic principle implies that each newly mentioned event shifts the narration time forward, thus providing a right boundary to the previously mentioned event. In such examples, a continuative reading is moreover prevented by the mention of other events between the ones which are repeated; otherwise the temporal boundary is explicitly given by lexical means such as after or next. The analysis of examples ((9) and (10a–c)) shows that the concept of iteration is easily expressible by the repertoire and the discourse organisation principles at work in the basic variety, that is, independently of verb morphology. Moreover again and its equivalents enhance textual cohesion through the anaphoric link they establish with the preceding co-text.

Verb morphology and the acquisition of temporal adverbs of contrast 

. Continuative and resultative items in postbasic varieties The acquisitional path and discourse functioning of other TACs is at first blush relatively less homogeneous than what we have seen for iterative items, but so are also postbasic varieties given that at this stage the learner system develops TLspecific features. Generally speaking the transition to postbasic varieties is marked by the development of a functional verb morphology, but this process presents some individual and crosslinguistic peculiarities. Learners of French and German L2 share remarkable similarities both in the evolution of their system into postbasic varieties and in the discourse functioning of TACs. In order to capture this (mutual) interaction, it is useful to look first at these learners and then at learners of English L2.

Learners of French and German L2 In both German and French L2 postbasic varieties, TACs such as déjà/schon are very frequently produced, but the contexts in which they occur do not always allow a temporal reading. Actually their acquisition seems to take place in 2 phases, according to the constituent in their scope and the semantic interpretation that the nature of this constituent imposes. Phase 1: déjà/schon + adverbials The initial occurrences of schon (Tino) and déjà (Bernarda, Alfonso) are mainly attested in utterances containing a temporal adverbial specifying either the frequency (11) or the position on the time axis (12) of the relevant event. (11) (about some dangerous friends, with whom Tino has gone out 3 times) schon drei mal mit ihm . . . mit waffen nur eine mal (I was) already 3 times with him . . . with weapons just once (Tino: cycle 2, rec. 27) In (11) the association of schon with an adverbial of frequency seems to rule out the pure temporal value of the analysed item. This appears most clearly with negation: the negative counterpart of temporal already is not yet. This negative transformation is not possible in (11): already three times is rather opposed to inferior quantities (just one or two). While in its temporal meaning the item has scope over the predicate, in (11) it has scope over the frequency adverbial three times and induces a scalar reading, that is, an evaluation of the relative quantity (Muller 1975; Löbner 1989). A similar remark can also be made for utterances like (12), which is representative of the initial occurrences of déjà.

 Sandra Benazzo

(12) je [fe] un stage de français. . . I do a training course of French déjà je [komãse] le vingt-sept de novembre already I start(ed) the 27th of november

(Bernarda: cycle 3, rec. 31)

Here, the association of déjà with an adverbial of position, indicating the reference time of the situation (the 27th of november), is quite bizarre: the presence of this TAC usually entails that the precise moment when the event has taken place is left undetermined, otherwise we once again obtain the scalar reading (Löbner 1989), so that already the 27th of november seems to be opposed to not just a few days ago. However in (12), and similar occurrences, a temporal interpretation of déjà cannot be completely excluded. Notice that at this stage tense marking with verb morphology is not systematic: in French L2, the auxiliated construction (aux – pp) is emerging in opposition to V for the present, but in many contexts the base form Vé is still attested (cf. komãse in (12)). As a consequence, the simultaneous presence of calendaric expression and of a resultative item could be considered as an attempt to mark temporal relations by several lexical means – adverbials, TACs and the lexical content of the verb commencer – in the absence of a systematic verb morphology. This is clearly the case in (13), where the past reference is indicated first by the lexical item avant, then by the calendaric expression 3rd March, which is in its turn further specified by the determination déjà passé. (13) (Bernarda about her daughter’s birthday) NS ça va être son anniversaire? is it going to be her birthday? oh non non [se] le dimanche avant. . . oh no no it is the sunday before le trois + le trois de mars déjà [pase] the 3rd of March already past

(Bernarda: cycle 3, rec. 32)

Alfonso, who is a bit more advanced than Bernarda, uses TACs (and especially déjà) in ambiguous contexts in order to place the relevant event in the past (14), although in other contexts without déjà, that occur at the same period, the verb is correctly inflected to mark passé composé. (14) (the NS has not understood whether the event related by AL refers to the past or to the future) NS tu y es allé + ou tu iras? have you gone there or will you? . . . je déjà je [ale] là bas oui . . . I already go there

Verb morphology and the acquisition of temporal adverbs of contrast 

et après je [ale] otra fois je [ire] otra fois and afterwards I go again I will go again

(Alfonso: cycle 1, rec. 19)

Contrary to French learners, Tino’s verbal opposition between simple and auxiliated forms in German L2 (V vs. Aux – geV) seems to be already well established at this stage, nevertheless schon appears mainly as a clear scalar particle (sometimes also as modal particle) or in association with other adverbials. Consequently it seems that, in spite of variable morphological marking on the verb, learners of German and French L2 go on anchoring their utterances in time through several means: verb morphology and temporal adverbials. Some scalar readings of déjà/ schon could be an epiphenomenon of grammaticalisation, but the status of these items is still ambiguous between scalar particle and aspectual operator during this first phase. Phase 2: déjà/schon + VP In a subsequent phase the frequency of time adverbials decreases, déjà and schon are more frequently attested in utterances where these items can mark the resultative or the semelfactive character of the event (example (15a–c)); whereas in the previous phase the relevant predicate had prototypically a past or perfect value,8 they now also appear in predicates with present tense and imperfective value (cf. (15c)). (15) a.

er hat nich gesehen das die/ + die zug ist schon weggefahrt he has not seen that the train has already left (Tino: cycle 3, rec. 33) b. tous les gens que j’ai déjà dit all the people I have already said (Alfonso: cycle 3, rec. 33) c. (ma fille) elle va déjà au lycée (my daughter) already goes to secondary school (Alfonso: cycle 3, rec. 33)

The transition between the first and the second phase is accompanied by further morphosyntactic development: –

–

in French L2 the distinction Aux – pp and V becomes more systematic, while learners start using periphrastic forms (such as verb aller + Vinf, or être en train de) and the imparfait of the copula; in German L2 we see the emergence of the Praeteritum for the copula and the modals, which is then opposed to Perfekt; imperfective aspect is sometimes expressed by the adverbial im Moment (Dietrich et al. 1995).

This evolution concerning the expression of temporality corresponds to a change in the scope of the analysed adverbs: in the first phase they affected principally a nominal adverbial phrase, now they affect more often the predicate and quantify over the time span indicated by the verb’s finite morphology.9

 Sandra Benazzo

As for continuative items, in both French and German L2, the additive particles noch and encore are now applied to time intervals (‘still’, as in (16)–(17)), although in French only one such occurrence is attested (17).10 (16) ich bin noch bose mit die andere I am still angry with the other

(Tino: cycle 2, rec. 27)

(17) NS tu as toujours peur de la mer? oui + encore oui are you always afraid of the sea? yes + still yes (Bernarda: cycle 3, rec. 32)

Learners of English L2 Already is only attested in the production of the most advanced learner, Lavinia, and with a relatively low number of occurrences (6 times in 2 cycles). Contrary to its French and German L2 equivalents, the first utterances with already do not contain time adverbials (example (18)), but this item appears at a stage where the learner already has a functional opposition between present and past, while aspectual distinctions (perfective, imperfective and perfect) are being elaborated and will be systematized shortly afterwards: (18) my landlord already + did + this kind of thing

(Lavinia: cycle 2, rec. 23)

In comparison to Tino, Bernarda and Alfonso, Lavinia’s occurrences of already correspond right from the beginning to other learners’ phase 2 and so does her verb morphology. Further development of aspectual distinctions is accompanied by the appearance of still and occurrences of already with predicates in the present tense (example (19a–b)). (19) a. Carlo is still working b. the course is full up already

(Lavinia: cycle 2, rec. 26) (Lavinia: cycle 3, rec. 31)

The less advanced learner Andrea shows a special path because of idiosyncracies which blur the picture. As it has already been pointed out, this progress beyond the basic variety only concerns the development of morphological tense oppositions, whereas aspectual distinctions (V / Ving or Ved / Aux – pp) are dysfunctional till the end of the observation period. At this stage again appears in contexts which imply a continuative reading (20). (20) (the learner is talking about the Italian town he comes from) when i had ten years twenty years ago . . . a lot of er woods NS and now? all the woods have gone? no there are + again but no like before (Andrea: cycle 2, rec. 24)

Verb morphology and the acquisition of temporal adverbs of contrast 

In the last recordings again is attested in its iterative function; meanwhile Andrea produces still, which, in alternation with stay, seems to be used as an idiosyncratic auxiliary, marking imperfective aspect (21a–b). (21) a.

(description of a picture where someone is buying apples) this man still to take some/ some apples into . . . his bag (Andrea: cycle3, rec. 31) b. (film retelling: the protagonist is calling his dentist) he stay to take an appointment with the dentist . . . . (Andrea: cycle3, rec. 32)

From what we have seen so far, there are some idiosyncratic features in Andrea’s development of postbasic variety verb morphology as well as in his use of TACs. Some of his overgeneralisations can probably be traced back to SL influence (Italian ancora corresponds to both ‘still’ and ‘again’, while Italian stare is both a lexical verb equivalent to ‘stay’ and an imperfective marker), but they are attested precisely at the time when the learner is elaborating functional tense oppositions on verb morphology.

. Comparison of the acquisitional paths and explanatory hypotheses The analysis has revealed commonalities as well as divergences in the acquisitional path of already/still: these items usually appear at postbasic variety level, but resultative already seems to be acquired in two phases (Bernarda, Alfonso, Tino) or in one step (Lavinia) according to the development of verb morphology; further morphological evolution seems to coincide with the emergence of continuative still (Lavinia, Bernarda, Tino), although other forms are also used for this function at a less advanced level (Andrea) but in an idiosyncratic way. The comparison of the three groups of learners suggests an explanation based on the development of verb morphology. From this perspective, the passage to postbasic varieties seems to be marked for all learners by at least two main developmental steps: –

–

first, a development of grammatical means encoding tense oppositions: the emergence of the auxiliary in French and German L2, while in English L2 simple forms for the present are opposed to the past of irregular verbs; later on, the appearance of grammatical means encoding aspect distinctions: periphrastic forms as Fr. aller + V, être en train de V, Engl. used to, going to or verbal morphemes as Germ. Präteritum, French imparfait for the copula, Engl. progressive form Ving and the distinction between Ved / Aux-Ved.

The correlation of this development with the observed sequence of TACs could be generalized as follows: a functional opposition of tense is necessary for the emer-

 Sandra Benazzo

gence and use of resultative already (which prototypically appears in past contexts), while a functional marking of aspect (perfective vs. imperfective) is required for the use of still (which needs imperfective aspect in order to get a continuative reading). This hypothesis perfectly accounts for the acquisitional path observed for example in Lavinia, however it does not explain the idiosyncrasies noticed in Andrea (continuative again or still/stay as markers of imperfectivity), nor many of the occurrences corresponding to the first phase of schon/déjà in Tino’s, Bernarda’s and Alfonso’s data (ambiguous status or attempts to mark past reference). However the exact opposite hypothesis provides a more powerful explanation: even if already/still generally appear in postbasic varieties, when verbal morphology is still ambiguous these items can be used as compensatory means contributing to tense (cf. déjà in French L2) or aspect (cf. still/again in Andrea’s production) marking of the predicate. When French L2 auxiliated forms are about to emerge in opposition to V but in competition with the basic form Ve, tense marking through verb morphology is not (yet) reliable. This is when Bernarda and Alfonso use déjà in alternation, or in association with, other temporal adverbials in order to signal past reference (cf. examples (12)–(14)). Later on, verb morphology of both French L2 learners, as well as Tino’s for German L2, is limited to one functional opposition: auxiliated forms are opposed to simple forms, but they are applied to all past contexts, while aspectual distinctions are expressed by adverbials or the lexical content of the verb. At this stage we can suppose that the presence vs. absence of schon/déjà allows these learners to distinguish perfect from perfective aspect, since this distinction is not marked by verb morphology alone (neither in the learner varieties nor in the standard TLs). Even if it is not always easy to state the exact function (scalar, temporal, aspectual) of the items analysed, the hypothesis of their compensatory role is clearly supported by the idiosyncracies also noticed in the less advanced learner of English L2, Andrea, who produces continuative again and still when he masters grammatical tense distinctions but not aspectual ones (examples (20)–(21)). A continuative item requires imperfective aspect, but in his production this condition is reversed: we infer imperfective aspect thanks to the presence of these items, which are thus used as markers of imperfectivity. On the contrary Lavinia quickly acquires the TL verb morphology which shows a very transparent form-function relationship as far as tense and aspect are concerned: the opposition between simple forms and progressive forms corresponds to the distinction of perfective/imperfective, while the perfect form encodes perfect aspect; these aspectual distinctions are available for the three tenses: present, past and future. In Lavinia’s production tense and aspect are rapidly specified by verb morphology, hence the potential compensatory role of already/still is not needed, they are immediately used with their standard temporal meaning.

Verb morphology and the acquisition of temporal adverbs of contrast 

On the basis of these statements, the unequal frequency of certain items (déjà/schon) as well as idiosyncratic uses (where still has the function of an imperfective auxiliary and again gives a continuative reading) could be due to the learners’ attempts to clarify an ambiguous verb morphology: –

–

as long as one verb form covering all present contexts is opposed to one form applying to all past contexts, verb morphology does not allow any aspectual distinctions for the specified tense: this aspectual specification is then expressed by TACs (déjà/schon idiosyncratically marking perfect; noch/still and Andrea’s again idiosyncratically marking imperfectivity); when the auxiliary is about to emerge, so that even morphological tense marking on the verb is not reliable, a resultative items can be used to signal ‘past’ reference, thus accounting for the initial occurrences of déjà in French L2.

We would like tentatively to extend this explanation to the sporadic, precocious presence of continuative noch in Angelina (1 occ.), déjà in Palmira (1 occ.) and maybe continuative again in Santo (cf. ex. (6c)), which are produced at the basic variety level. Even though such low numbers do not allow any precise statements, their presence suggests that learners fossilized as for grammatical structures can extend the communicative potential of their repertoire by acquiring compensatory lexical items, which however allow different possible interpretations, as in Santo’s utterance given in (6c) (and repeated here), and in Palmira’s utterance following it.11 (6) c.

(Santo is waiting for his driving licence) maybe i wait again another month I keep on waiting? Iwait one more month? Oce again,Iwait one month (Santo: cycle 1, rec. 16)

(22) (Palmira says she has headache) NS maintenant? now? déja déja hier soir eh *dolor* already yesterday evening pain = already yesterday? from yesterday ? also yesterday? (Palmira: cycle 3, rec. 36) The interrelation established between underdetermined verb morphology and presence of TACs is confirmed by other studies. Carroll and von Stutterheim’s (1997) analysis of temporal reference in German and English narratives reveals the speaker’s preference for different linguistic means according to the language in question: temporal adverbs in German and aspectual morphology in English. The speaker’s choice would reflect the linguistic ressources available in the respective language: the English system allows a rich repertoire of temporal–aspectual dis-

 Sandra Benazzo

tinctions through verb morphology, which are not possible through the German verb system, so that German speakers resort more often to temporal adverbs. Dankova (1997: 189), who studied narratives produced in Esperanto by Italian, French and Russian speakers, has stated a similar connection: “plus la morphologie verbale est riche, moins les expressions adverbiales de temps sont nécessaires . . . si les italophones expriment des différentiations aspectuelles par des moyens verbaux, les francophones et surtout les russophones”, who show a less varied verb morphology, “sont amenés à utiliser les adverbes de nature de transition et de contraste”. To paraphrase briefly: speakers who have acquired a rich verb morphology in Esperanto, use few adverbs of contrast; while speakers who show a less rich verb morphology use more adverbs of contrast.

. Conclusions The proposed analysis has shown a similar order of appearance and use of scopal items at a given acquisitional stage of L2 in the production of learners of different TLs. In order to generalize the statements of the previous sections, we integrate the acquisition of TACs into the developmental sequence for the expression of temporal relations proposed by Dietrich et al. (1995), as in Figure 1. In prebasic varieties no TACs are attested, although there may be other scopal items (additive/restrictive particles). At the basic variety temporal relations are mainly explicited by lexical means, which allow learners to mark event boundaries or to specify “some time span, its position on the time axis, its duration and (if iterated) its frequency” (Dietrich et al. 1995: 268). From this stage on appear iterative items: from a discourse perspective, they establish an anaphoric link with a previously mentioned event; from a temporal point of view they make reference to another occurrence of the same event at a previous non adjacent time interval, from which the actual one is separated by mention of other events (PNO) or by

Figure 1. The developmental sequence for the expression of temporal relations

Verb morphology and the acquisition of temporal adverbs of contrast 

the presence of lexical elements such as after or next. Iterativity perfectly fits the organizing principles at work at basic variety level. In postbasic varieties the main change is represented by the emergence of finite verb morphology. Learners develop at least one verb form for the present, opposed to one verb form for the past; learners may then elaborate grammaticalized aspectual oppositions. This development has an important consequence for discourse organisation: verb morphology can signal by itself whether the temporal reference is valid for the present or for the past, a function that was previously accomplished by temporal adverbs or inferred by default only. At this stage items such as already and still can be used to relate and contrast different time intervals of the same event. The association of these adverbs with a functional verb morphology is actually necessary to get their standard temporal meaning: already marks the transition from a negative to a positive phase of a state or an event, while still marks its continuation from a previous undetermined time interval. In both cases finite verb morphology indicates which is the actual time span of the event talked about (Klein 1994’s Topic Time), while TACs make reference to alternative (previous) time spans of the same event. The transition between the two varieties can be marked by an intermediate phase where morphological variation is not yet reliable or systematic, so that temporal relations are still mainly expressed by lexical means. In this phase already/still have an ambiguous status: they can be used as scalar particles or as compensatory means contributing to the marking of tense (already = past) or aspect (already = perfect; still = imperfective), but not as adverbs of contrast. The rare occurrences of these particles attested in the production of learners who do not go beyond the basic variety level can also be accounted for in this way. Coming back to the initial questions, we have thus described the emergence and use of TACs in relation to the development of learner varieties and their place among other devices to express temporal relations, but what are the reasons explaining this developmental sequence? Theoretically nothing prevents learners from using all of these adverbs from the initial stages of L2 acquisition, as they do with other adverbials. But we have seen that this does not happen. On the contrary the items we have analysed appear in a fixed order and the particle repertoire gets enriched in parallel with the development of the learner’s system towards the TL morphology. The sequence of appearance and use of the items in question seems indeed to be determined on the one hand by the development of the learner variety, given that this is the systematic feature common to the 3 groups of learners, and on the other hand by their discourse functions. This correlation becomes clearer by considering not only TACs but also other scopal items and the relative constituent they have scope over:

 Sandra Benazzo

–

–

–

additive/restrictive particles, attested from prebasic varieties onwards, quantify over the NP-referents or adjectives which are typical for the repertoire of this stage; iterative items, produced from the basic variety level onwards, quantify over tokens of events which have to be bounded and the repertoire of the basic variety allows the expression of bounded events; resultative/continuative items relate and contrast time intervals of a single event and, in order to do so, need a functional verb morphology to specify the relevant time interval.

This correlation provides an alternative explanation for the late appearance of certain items. The reasons for the observed acquisitional sequence are to be found in the constraints governing the learner system at a given time and in the discourse functioning of the items in question, rather than in the conceptual complexity of TACs: internal factors concerning the grammaticalization process observed in learner production seem to be better candidates than cognitive factors in explaining the acquisition of TACs.

Notes * A first version of this paper was presented at the Euroconference on “The Dynamics of Learner Varieties” in Sant-Feliu (Spain), October 2000. I am very grateful to Clive Perdue for many comments and valuable suggestions on preliminary versions. . There are many different interpretations of the meaning of TACs. Some authors consider them in terms of the speaker’s expectations: already/still would signal that the relative event has taken place earlier or later than expected (cf. Hoepelman & Roeher 1980; Vet 1980). Others prefer to consider these items as aspectual operators, while such expectations are treated as contextual pragmatic inferences, which are independent of their basic (temporal) meaning (cf. Löbner 1989; Borillo 1984 and Klein 1994 among others). In the following we adhere to the second view. . In other contexts, as in John has already been to Spain, this item implies a ‘semelfactive’ reading: it means that the relative event ‘John’s being to Spain’ has taken place at least one time before the time of utterance. In this case the negation of already is not not yet, but never. . For a more precise definition of these aspectual notions, cf. Klein (1994). . Learner examples present the following conventions: *..* enclose passages from SL; [.. ] contain the phonetic transcription of ambiguous verb forms; ... signal omitted passages; + means a short pause; NS introduces native speaker turns; scopal items are given in bold. . In comparison to English and German L2, iterative items are quite rare in the production of French L2 learners. A possible explanation is that in Romance languages iterativity is also frequently expressed by the verbal prefix re- like in repasser for come again or redescendre for go down again. Spanish-speaking learners of French seemingly rely more on lexical verb

Verb morphology and the acquisition of temporal adverbs of contrast 

aspect rather than on specific adverbs, as can be seen in the following extracts. In order to get the interlocutor’s question repeated, the Italian learner of German Tino says “bitte nochmal” ‘please (say it) again’; while the Spanish-speaking learner of French Alfonso says : vous pouvez répéter, s’il vous plaît? (‘could you repeat, please’?). . A typical example is je ne sais plus. For more details on the acquisition of negative items by learners of English and French L2 see Giuliano (this volume), and for learners of German L2 see Becker and Dietrich (1996). . This is a possible explanation of some idiosyncratic uses mentioned in Tables 2–4: certain learners take advantage of the structural similarity between additive and iterative items, using iterative particles instead of additive ones (e.g. again in Santo) and additive particles instead of iterative ones (e.g. auch in Angelina). For more details about the anaphoric linking established by additive particles see Watorek and Perdue (1999) and Benazzo (2000). . Some authors consider these occurrences in terms of iterative meaning (Gosselin 1996), as opposed to durative meaning (already/still + imperfective aspect). . Besides the detailed occurrences, we also find other non-temporal occurrences of déjà/schon, which are used for example as modal particles (e.g. ich kann schon sagen). Although interesting in themselves – since they imply a diversification in the functions of schon/déjà – they will not be discussed here because they are beyond the scope of this paper. For more details see Benazzo (2000). . The tiny number of continuative encore is probably due to SL influence: Spanishspeaking learners of French L2 seem to resort more often to verbs like continuer (‘go on’, continuar in Spanish), instead of using specific adverbs (cf. note 6 for iterative items). . There seems to be a slightly different acquisitional path for learners progressing beyond the basic variety and those fossilized at this level: in the former case, learners are more oriented towards TL structures and use already/still contrastively once they have reached a postbasic level; in the latter case, they fossilize as for grammatical structures, but progress further in the acquisition of scopal items by attributing new functions to the repertoire they already have (f.ex. again/noch with continuative meaning).

References Becker, A. & Dietrich, R. (1996). The Acquisition of Scope in L2 German. Zeitschrift für Literaturwissenschaft und Linguistik, 104, 115–140. Benazzo, S. (2000). L’acquisition de particules de portée en français, anglais et allemand L2. Etudes longitudinales comparées. Unpublished doctorate thesis, Université Paris VIII / Freie Universität Berlin. Borillo, A. (1984). La négation et les modifieurs temporels: une fois de plus ‘encore’. Langue Française, 62, 37–58. Carrol, M. & Stutterheim, Ch. von (1997). Relations entre grammaticalisation et conceptualisation et implications sur l’acquisition d’une langue étrangère. AILE, 9, 83–115.

 Sandra Benazzo

Dankova, N. (1997). Temporalité en espéranto. Etude du transfert. Unpublished doctorate thesis, Université Paris VIII. Dietrich, R., Klein W., & Noyau, C. (Eds.). (1995). The Acquisition of Temporality in Second Language. Amsterdam: John Benjamins. Franckel, J.-J. (1989). Etude de quelques marqueurs aspectuels du français. Genève-Paris: Droz. Giuliano, P. (2000). L’acquisition et l’expression des fonctions négatives en français et en anglais comme langues secondes. Unpublished doctorate thesis, Université Paris VIII / Università di Pavia. Gosselin, L. (1996). Sémantique de la temporalité en français. Paris: Duculot. Hoepelman, J. & Rohrer, Ch. (1980). ‘Déjà’ et ‘encore’ et les temps du passé du français. In J. David et R. Martin (Éd.), La notion d’aspect. Recherches linguistiques (pp. 119–144). Etudes publiées par le Centre d’Analyse de l’Université de Metz, Paris: Klinksieck. Klein, W. (1994). Time in Language. London: Routledge. Klein, W. & Perdue, C. (1992). Utterance Structure: Developing Grammars Again. Amsterdam: Benjamins. Klein, W. & Perdue, C. (1997). The Basic Variety. Second Language Research, 13(4), 301–347. König, E. (1977). Temporal and non-temporal uses of ‘noch’ and ‘schon’ in German. Linguistics and Philosophy, 1, 173–198. König, E. (1991). The Meaning of Focus Particles: A Comparative Perspective. London: Routledge. Löbner, S. (1989). German ‘schon-erst–noch’: an integrated analysis. Linguistics and Philosophy, 12, 167–212. Muller, C. (1975). Remarques syntactico-sémantiques sur certains adverbes de temps. Le Français Moderne, 43, 12–38. Perdue, C. (Ed.). (1993). Adult Language Acquisition: crosslinguistic perspectives. Cambridge: Cambridge University Press. Perdue, C., Benazzo, S., & Giuliano, P. (2002). When finiteness gets marked: the relation between morpho-syntactic development and use of scopal items in adult language acquisition. Linguistics, 40, 849–890. Sato, C. (1990). The Syntax of Conversation in Interlanguage Development. Tübingen: Gunther Narr. Starren, M. & Van Hout, R. (1996). Temporality in learner discourse: What temporal adverbials can and what they cannot express. Zeitschrift für Literaturwissenschaft und Linguistik, 104, 35–50. Starren, M. (2001). The Second Time: The Acquisition of Temporality in Dutch and French as a Second Language. Doctoral thesis, Utrecht: LOT. Vet, C. (1980). Les adverbes présuppositionnels. In Temps, aspect et adverbes de temps en français contemporain (pp. 195–209). Genève: Droz. Watorek, M. & Perdue, C. (1999). Additive particles and focus: Observations from learner and native-speaker production. Linguistics, 37(2), 297–323.

JB[v.20020404] Prn:26/06/2003; 10:29

F: SIB2608.tex / p.1 (35-102)

Merging scope particles Word order variation and the acquisition of aussi and ook in a bilingual context* Aafke Hulk

.

Introduction

The distribution of scope particles such as the French aussi and the Dutch ook is known to interact with the informational, the phonological and the syntactic structure of the utterance. In language acquisition the learning task therefore consists in having to acquire the principles of information structural organisation in addition to the linguistic devices, and the functional interrelation between these two systems. As for the syntactic properties of ook/aussi, in the acquisition literature (i.e. Penner et al. 1995) these elements have been claimed to function as pre-cursors of functional categories of the verbal domain. Semantically, they are “bridging” elements, relating an utterance to the preceding discourse. Now it is well known that in comprehension tests children generally have problems with the integration of elements into the discourse, i.e. Karmiloff-Smith (1981) who argues that young children treat each sentence as a separate discourse unit. However, whereas other “bridging” elements may be rare or difficult to interpret in spontaneous production data, elements such as aussi/ook appear early and frequently in child production data. Therefore the study of their development (in production) is important for our knowledge about the early acquisition of phenomena involving the interface between syntax and pragmatics, complementing the results of the comprehension studies on other, similar elements. In this paper we will present a case study of the production and development of aussi and ook in the bilingual French/Dutch girl Anouk. We will compare her data to data from two monolingual children, the Dutch girl Laura and the French boy Philippe, both available on Childes (MacWhinney & Snow 1985).

JB[v.20020404] Prn:26/06/2003; 10:29

F: SIB2608.tex / p.2 (102-146)

 Aafke Hulk

Consider the following dialogue between the bilingual girl Anouk, age 2;7.5 (2 years, 7 months and 5 days), and her French mother Christiane: (1) Ch: il est cassé celui-là ‘he is broken that one’ A: aussi cassé ‘also broken’ Ch: pas aussi cassé, Anouk, il est cassé aussi ‘not also broken, A, he is broken also’ A: celle-là aussi cassé ‘that one also broken’ Ch: il est cassé aussi ‘he is broken also’ Apparently the mother does not approve of the word order Anouk is using with respect to the position of aussi. Is this the well known situation where a parent tries to correct the “child-French” of her daughter, who, in this case, is struggling with a specific particle? Or is it a francophone mother correcting the Dutch-like French her bilingual daughter is speaking? In order to answer these questions we will first briefly consider the syntax and semantics of the particles aussi and ook in respectively French and Dutch. Then we will examine the acquisition of such particles by monolingual children and finally we will come back to the bilingual Anouk and look at her use of aussi/ook in more detail.

. Semantics and pragmatics of aussi/ook When aussi/ook occurs in a sentence, this sentence is automatically linked to the preceding discourse in a specific way: the elements in the scope of aussi/ook are presupposed to be in a paradigmatic relation with structurally and conceptually similar elements in other sentences. A sentence such as the following (2) Marie aussi pleure / Ook Marie huilt ‘Marie also cries’ expresses the presupposition that others are also crying; it adds Marie to the (list of) persons who are said to cry. Therefore, elements such as aussi/ook are sometimes called additive particles. It is important to distinguish the potential scope of the particle and the element which is actually semantically affected. Whereas the scope of the particle can be characterized in syntactic terms, i.e. as the c-command domain, to determine which of the elements in the scope is/are actually affected, we need more than syntax1 (cf. Watorek & Perdue 1999): the preceding context

JB[v.20020404] Prn:26/06/2003; 10:29

F: SIB2608.tex / p.3 (146-197)

Merging scope particles 

has to be taken into account. Elements such as aussi/ook have a discourse organisational function and, therefore, are sometimes called discourse particles. In the following sentence both the subject Marie and the verb pleure/huilt are in the scope of aussi/ook; we need the preceeding discourse to determine which element is actually affected: (3) Marie pleure aussi / Marie huilt ook ‘Marie cries also’ The affected element often expresses new information and can be qualified as the (informational) focus of the sentence. Therefore particles such as ook/aussi are sometimes called focus particles. However several authors (Dimroth 2000; von Stechow 2000) have shown for the German particle auch that in some cases, the affected element can better be qualified as a (contrastive) topic: auch then links two topics for which some state of affairs is claimed to hold. Clearly this also holds for aussi and ook: in the appropriate context, for example, where several people are crying and Marie is watching them, Marie in sentences such as (2) and (3) above, is the topic of the sentence and the semantically affected element and aussi/ook functions as a topic particle. In another context (3), but not (2), can be used to express the new information that besides doing other things, Marie is also crying: in that case pleure/huilt is the affected element and the informational focus of the sentence, and aussi/ook functions as a focus particle. In this article we will use the neutral term discourse particle.

. The syntax of aussi/ook Above we saw that both aussi and ook have the same semantics/pragmatics: they are scope-taking discourse particles which can affect either the focus or the topic of the sentence. Here we will see that they differ as far as the syntax is concerned: they do not appear in the same positions in the sentence. This is of course related to general word order differences between Dutch and French. Dutch is an V2 language with underlying OV (left branching) word order, whereas French is VO (right branching) and non-V2. We will not discuss the syntax and its interaction with the semantics and phonology in any detail here.2 We will just assume that particles such as aussi/ook can be generated either adjoined to (the constituent containing) the affected element or in a Functional Projection above VP (a ScopePhrase or an AffectPhrase).

JB[v.20020404] Prn:26/06/2003; 10:29

F: SIB2608.tex / p.4 (197-271)

 Aafke Hulk

. French In French aussi can appear in three different positions (cf. Perrin-Naffakh 1996; Nolke 1993):3 (i) sentence final position (4) Marie travaille dans le jardin aussi ‘Marie works in the garden also’ (5) Marie devient paresseuse aussi ‘Marie gets lazy also’ In that case aussi can affect either one of the preceding elements separately or a combination of the elements inside the VP or even the whole sentence. If aussi only affects the constituent to its left, i.e. dans le jardin in (4), we will assume it is right adjoined to this constituent. If it is not adjacent to the element it affects, it probably occupies a rather high position. One possibility would be to assume that, in that case, it is right-adjoined to the sentence (IP or CP).4 (ii) immediately to the right of the XP in first position This XP can be the subject of the clause,5 as illustrated in (6): (6) Marie aussi travaille dans le jardin ‘Marie also works in the garden’ In such cases, the part of the sentence following aussi will often be omitted: (7) Pierre travaille dans le jardin. Marie aussi. ‘Pierre works in the garden. Marie also’ However, the XP in first position can also be another fronted constituent, which precedes the subject,6 as illustrated in (8) and (9): (8) sur le lit aussi il y a une dame ‘on the bed also there is a lady’ (9) ça aussi j´adore ‘that also I like’ In these sentences it is only the XP in first position which is in the scope of aussi. We assume that here aussi is right adjoined to this XP.7 (iii) between the finite verb and what follows:8 (10) Marie a aussi dormi ‘Marie has also slept’

JB[v.20020404] Prn:26/06/2003; 10:29

F: SIB2608.tex / p.5 (271-338)

Merging scope particles

(11) Marie parle aussi à son frère ‘Marie speaks also to her brother’ (12) Marie travaillera aussi dans le jardin ‘Marie will work also in the garden’ In sentences such as (10)–(12) the whole sentence is in the scope of aussi: depending on the context (and the accentuation) either the subject, the finite verb, the other element or a combination of the latter two or the whole sentence can be actually semantically affected. These elements appear not only in a position to the left of aussi (as was the case above) but also to the right. This faces us with the problem in which position aussi is generated here. Above we considered the possibility that aussi was right adjoined to the element in its scope. Here that seems less plausible, unless we right adjoin aussi to INFL (the finite verb). However, certain adverbs, such as probablement, can appear between the finite verb and aussi in this construction:9 (13) Marie a probablement aussi attrapé le virus ‘Marie has probably also got the virus’ Therefore adjunction to INFL is not very likely.10 Another, more plausible option would be to assume that there is a functional projection ScopePhrase between IP and VP with aussi in its head position.11

. Dutch In Dutch ook can appear in two positions: (i) in sentence initial position Here we have to distinguish two different possibilities: either ook is immediately followed by the finite verb, as in (14), or ook is followed by an XP. This XP can be the subject, as in (15), but it can also have any other function, e.g. direct object as in (16): (14) Ook heeft Jan een virus opgelopen ‘also has Jan a virus got’ (15) Ook Jan heeft een virus opgelopen ‘also Jan has a virus got’ (16) Ook aardbeien lust hij graag ‘also strawberries likes he well’ In (14) ook has scope over the entire sentence. One could suggest it is head of a ScopePhrase with the whole CP as its complement.12 In (15) and (16) ook has only



JB[v.20020404] Prn:26/06/2003; 10:29

F: SIB2608.tex / p.6 (338-407)

 Aafke Hulk

scope over Jan/aardbeien, which bears a pitch accent; the interpretation is contrastive. Since Dutch is strictly V2, i.e. only one constituent can precede the finite verb in root clauses, ook XP has to form one constituent here. Therefore it seems plausible to assume that ook is left adjoined to XP in these cases. (ii) between the finite verb and what follows This is the most common position for ook: (17) Marie heeft ook geslapen ‘Marie has also slept’ (18) Marie gaf ook een boekje aan Piet ‘Marie gave also a book to Piet’ Just as aussi in French, ook can affect elements either to its right or to its left (depending on the context and the accent). Moreover it does not have to be adjacent to the finite verb: certain adverbs (19), the inverted subject (20) or a scrambled object (21) may intervene between the finite verb and ook: (19) Marie heeft waarschijnlijk ook geslapen ‘Marie has probably also slept’ (20) Dat heeft Marie ook gezegd ‘that has Marie also said’ (21) Marie heeft dat boek ook aan Piet gegeven ‘Marie has that book also to Piet given’ Since Dutch is a left-branching language and ook generally precedes the element in its scope, it would be possible to assume here that ook is left-adjoined to VP. It is also possible to assume that, as in French, there is a functional projection ScopePhrase just above VP. Plausibly, both possibilities co-exist for ook in Dutch: adjunction to VP, in the case the affected element is (inside) the VP, and head of a Scope Phrase in all other cases. The sentence final position is never possible as a base-position for ook, contrary to what we saw for aussi: (22) *Marie heeft geslapen ook ‘Marie has slept also’ However, it is possible to find ook in sentence final position, if the object has been scrambled, as in (23), or wh-moved, as in (24), across ook:13 (23) Marie wil dati ook ti ‘Marie wants that also’

JB[v.20020404] Prn:26/06/2003; 10:29

F: SIB2608.tex / p.7 (407-462)

Merging scope particles 

(24) Wati wil Marie ook ti ? ‘what wants Marie also’ Summarizing, we saw that French and Dutch share the possibility to have ook/aussi in sentence internal position just above VP. In Dutch ook in that case is either leftadjoined to VP or in a FP above VP; in French only the latter is possible. French and Dutch differ in two other respects: (1) French freely (and frequently) allows aussi in sentence final position, which is possible only in very restricted cases in Dutch, i.e. when another element has moved over it, (2) Dutch allows ook in sentence first position, either immediately preceding the finite verb, or forming a constituent with the sentence initial XP to its right. In French, the only option for aussi in the left periphery is to follow the constituent (which it affects) in first position.

. The acquisition of aussi/ook . Dutch In the acquisition literature it has been argued that the German element auch and the Dutch ook fulfil a pioneering function for syntactic structure building, since they provide the basis for the first functional “Affect”projection above VP (Penner et al.1995; Tracy 1995; Powers 1996). Let us take a closer look at how children build their sentences.14 We assume, following Roeper (1995) and Powers (op.cit.) among others, that children start out building utterances by merging together two lexical elements. These elements do not yet have their adult properties: they can be incomplete or underspecified, both morpho-syntactically and semantically. As has been already argued by Brain (1976) children make novel use of adult lexical items. For example: they use “more X” or “no X” to express a whole array of meanings. Similarly, van Kampen (1997) observes that Dutch children initially overgeneralize the use of ook: for them ook often just marks the sentence with a strong affirmative meaning (cf. also Dimroth et al., this volume). Roeper calls such novel projections created by the child: Unique Maximal Projections (UMPs): (25)

UMP more

X(P)

UMP ook

X(P)

In the course of the acquisition path the elements forming the UMP become more specified and thus compatible with the set of adult maximal projections. Powers claims that ook is first merged to an XP, second it is used as a novel modality marker

JB[v.20020404] Prn:26/06/2003; 10:29

F: SIB2608.tex / p.8 (462-537)

 Aafke Hulk

adjoined to VP. Finally when it acquires the feature [+mod/+v] it is reanalysed as a (functional) head.15 Penner et al (op.cit.) observe for the acquisition of auch by German children that after a first period where auch appears in an unstable, peripheral position (26), in a second stage it appears utterance internally and is in complementary distribution with finite verbs (27) and (28): (26) Stephanie auch / auch Stephanie ‘Stephanie also / also Stephanie’ (27) Stephanie auch nase ‘Stephanie also nose’ (28) ich auch Brot haben ‘I also bread have’ The same pattern has been found for ook in child Dutch ( Dimroth et al. op.cit; van Kampen op.cit; Powers op.cit.). The following example is the first occurrence of ook in Laura:16 (29) Laura ook ‘Laura also’

Lau 1;10.10

We could analyze this utterance as an UMP, where two lexical elements have merged. In three word utterances, ook is often used in a context in which adults would use the modal auxiliary verb willen or moeten: (30) papa ook eten ‘daddy also eat’

Laura 2;0.28

(31) ik ook sokken uit ‘I also socks off ’

Laura 2;0.29

(32) ik ook peer ‘I also pear’

Laura 2;1.2

Three word utterances with ook in second position can be generated in two ways: –

–

by Merging/left-adjoining: For an utterance such as (30) this would imply first merging ook with eten into an UMP, let us call it an OokPhrase, and then merging papa with this OokPhrase into a regular child-Small Clause, as illustrated in (33); or by Merge + Move: For (30) this would imply first merging ook with the VP papa eten and then moving papa out of the VP and adjoining it to the left of ook, as illustrated in (34).

JB[v.20020404] Prn:26/06/2003; 10:29

F: SIB2608.tex / p.9 (537-591)

Merging scope particles 

(33)

SC Papa

OokP ook

eten

FP

(34) ook

FP papa

VP papa

eten

FP ook

VP eten

It seems plausible to assume that (33) represents an early stage when ook is still underspecified for its features and that (34) arises when ook has been reanalysed as a new modality marker with functional features. For some time the two structures may co-exist. When we consider the distribution of ook in the first 9 files of Laura (age 1;9.18 to 2;0.19), we get the following picture: her two word utterances with ook are mostly of the type XP-ook and less frequently of the type ook-XP; about half of her utterances with ook (52%) are three word utterances of the type in (30), with ook in between two other elements.

. French The acquisition of particles such as aussi by monolingual French children did not get much attention in the literature. There is just one unpublished paper by Gayraud (1997) who looked at data from Grégoire and Philippe, available at CHILDES. She found that, between age 1;9.18 and 2;5.27, Grégoire hardly uses aussi at all and that in all [8] cases it appears in second position, following a noun or a pronoun, indicating a change of topic. As for Philippe, between age 2;1.19 and 2;11.21, she found some development in the use of aussi: before age 2;3 aussi appeared in initial or preferably in second position, after that age the tendacy was for aussi to appear in postverbal or final position, with a clear preference for the final position. We also looked at Philippe’s data and studied the first twenty recordings (age 2;1.19–2;8.15). In Philippe’s data aussi is found from the first recording onwards. The first five occurrences all are of the form NP aussi: (35) Myriam aussi? ‘Myriam also’

JB[v.20020404] Prn:26/06/2003; 10:29

F: SIB2608.tex / p.10 (591-650)

 Aafke Hulk

In Philippe age 2;2.10 we find the first example of aussi in final position of an utterance with a (finite) verb; however this is a repetition (36) Ecrit bien aussi ‘writes well also’ The next example with aussi in final position (in an utterance containing a verb) is found in Philippe age 2;2.26: (37) y avait des belles autos aussi ‘there were beautiful cars also’ This utterance is produced as a reaction to the following utterance by his father, indicating that aussi in (37) is adjacent to the affected element (des belles autos) which has focus function here. (38) il y avait des drapeaux ‘there were flags’ The right peripheral position is the most frequent one in Philippe’s data. In that position aussi mostly functions as a focus particle. In (39) the verb expresses the new information: (39) ça pique aussi ‘that stings also’

Phi 2;3.21

The first example of aussi in an utterance internal position, is found in Philippe age 2;3.21: (40) là aussi cassé17 ‘there/that one also broken’ This example is particularly interesting because in the introduction of this paper we saw a similar example (1) produced by the bilingual girl Anouk which was not accepted by her francophone mother. Apparently monolingual French children do produce similar utterances with aussi in internal position. What could be the structure of such an utterance? Is it structurally ambiguous just as the Dutch three word utterances with ook in internal position (cf. (30))? This is indeed plausible since (40) could correspond to two different utterances in adult French, one where aussi is right adjoined to celui-là (41) and one where aussi is in a (FP) position in between the finite verb and cassé in the VP (42): (41) celui-là aussi (il est) cassé ‘that one too (it is) broken’

JB[v.20020404] Prn:26/06/2003; 10:29

F: SIB2608.tex / p.11 (650-707)

Merging scope particles 

(42) celui-là (il est) aussi cassé ‘that one it is also broken’ Let us assume then that the child-utterance (40) is structurally ambiguous: it can arise either just by Merge (or right-adjoin) or by a combination of Merge + Move. The first case, merging aussi with là into an UMP, let us call it an “aussiPhrase”, and then merging cassé and this “aussiPhrase” into a child Small Clause is exemplified in (43): (43)

SC aussiP là

cassé

aussi

The second possibility to create (40) is to first merge the VP [là cassé] to aussi and then moving là out of the VP across aussi, as illustrated in (44): FP

(44) aussi

FP lài

VP là

cassé

FP aussi

VP ti

cassé

Philippe is talking about broken pencils when uttering (40). Therefore, aussi in (40) affects là, and functions as a topic particle. In such case, (43) is the most plausible structure: aussi is adjacent to the element which it affects. Moreover, (44) would correspond to the adult (42) which would be avoided in adult French, since, according to Perrin-Naffakh (1996), in sentences with aussi between a copula (est) and an adjectival predicate (cassé) a confusion could arise with sentences where we find another element aussi expressing a comparative degree (cf. Note 8). This might be the reason why Anouk’s mother did not like her producing the word order in (1). It is not clear though whether (French) children are already aware of this possible confusion. In Philippe’s later data we find a lot of utterances with an internal aussi, where it is unambiguously adjoined to the XP in first position, which it affects: (45) moi aussi je vais écrire avec Madeleine ‘me too I go write with M’

Phi 2;6.13

(46) là aussi c’est chez mami ‘there too it is at granny’s’

Phi 2;7.25

JB[v.20020404] Prn:26/06/2003; 10:29

F: SIB2608.tex / p.12 (707-778)

 Aafke Hulk

(47) du beurre aussi dessus le croissant ‘butter also on the croissant’

Phi 2;8.15

(48) à Orsay aussi y a le chateau de Marc ‘in Orsay also is the castle of M’ In the first 20 files of Philippe we do not find any examples with an internal aussi which unambiguously occupies a position in the FP above VP. Such utterances appear only from age 2;8.29 onwards: (49) on le déchire aussi pour l’enlever ‘one it tears also to it take off ’

Phi 2;8.29

(50) en a aussi dans celui-là ‘of it is also in that one’

Phi 2;9.15

When we summarize the positions of aussi in the first 20 files of Philippe, we get the following picture: the utterance final position is the most frequent (66%), the left peripheral position (utterance first) is rare (2%) and in one third (31%) of the cases we find aussi in internal/second position as in (40). It is difficult to compare these percentages to what we have seen for the Dutch child Laura, since the period covered by the data studied here is much larger for Philippe than for Laura. Nevertheless we can say that whereas in Laura utterances final ook appears mainly in two word utterances of the type XP-ook , and never in three word utterances, utterance final aussi in Philippe appears also and frequently in multi-word utterances. This corresponds to what we know about the adult languages (cf. Section 2).

. Anouk: Aussi and ook in bilingual acquisition In the last decade, most studies of bilingual language acquisition have shown that bilingual children are able to differentiate between their two languages from early on (Genesee 1989; De Houwer 1990; Meisel 1989). Recently, the separate language hypothesis has been further refined: the development of two languages in a bilingual child may be largely autonomous, but this does not exclude the possibility that there can be influence from one language on the other (Döpke 1998; GawlitzekMaiwald & Tracy 1996; Hulk 1997; Müller, 1998; Müller, Hulk, & Jakubowicz 1999). In such cases we may speak of “cross-linguistic influence”. Cross-linguistic influence then is not to be taken as mixing or fusion, but it could show up as facilitation/acceleration, delay, or transfer (Paradis & Genesee 1997). In studying bilingual language acquisition, we observe the emergence of grammars of two languages at the moment of creation, when they are in close contact with each other. The bilingual child may attempt to make short cuts and allow strategies from

JB[v.20020404] Prn:26/06/2003; 10:29

F: SIB2608.tex / p.13 (778-836)

Merging scope particles 

one language into the other. Plausibly, such “short cuts” are taken when the child has to cope with problematic input. The interesting question is to find out what is this problematic input, i.e. which parts of the grammar are sensitive to such cross-linguistic influence and why this should be so. Hulk (2000), Hulk and Müller (2000), Müller and Hulk (2001) have observed syntactic cross-linguistic influence in particular in phenomena involving the interface between syntax and pragmatics: e.g. they found more object fronting and omissions in Anouk’s French than in monolingual children, and a delay in the acquisition of subject clitics and in the disappearance of root infinitives. In all these cases, a construction/word order which was possible, but not so frequent in monolingual child language, was used more frequently and/or for a longer period by the bilingual child. What do we expect with respect to the acquisition of aussi/ook? These elements clearly involve the interface between syntax and pragmatics, both in Dutch and in French. Although French and Dutch differ as for the positions in which aussi and ook can appear, they also show some overlap: in both languages these elements can appear in a so-called internal position, which was identified as a position inside a Functional Projection above VP. We also saw that in a certain developmental stage both Dutch and French monolingual children use aussi/ook in the internal position of three word utterances without a finite verb (cf. also Jordens, to appear). These utterances can be analysed as the precursors of the adult utterances with a finite verb and a fully specified FP containing aussi/ook.

. Aussi in Anouk’s French Anouk has been living in Amsterdam from birth and is being brought up bilingually by her French mother who speaks only French to her and her Dutch father, who speaks only Dutch to her, following the strategy of “une personne, une langue”. Anouk’s mother understands Dutch and speaks it reasonably well. Anouk’s father understands French but speaks it poorly. From about 6 months of age onwards Anouk attended a Dutch kindergarten for three days a week. The corpus was collected by making audio recordings of both languages. The recordings were made by the mother for French, and by the father – and a native Dutch student – for Dutch, at approximately three weeks intervals, starting at age 2;3.13, when Anouk first produced intelligible utterances of more that one word, until the age of 3;10.7. Unfortunately, the Dutch data were not collected as frequently as planned, and therefore we have fewer data on Dutch than on French. All data were transcribed by a native Dutch student of French in the CHAT-format. No phonetic transcription was made. In the first two files there are no utterances containing aussi. The first occurrence of aussi appears at age 2;4.17: aussi it is adjacent to the element express-

JB[v.20020404] Prn:26/06/2003; 10:29

F: SIB2608.tex / p.14 (836-886)

 Aafke Hulk

ing new information, but the positioning is still unstable, as illustrated in the following dialogue: (51) Mother: Anouk: Mother: Anouk: Mother: Anouk:

qu’est-ce que tu vas faire, un autre petit chien? ‘what you go make, another little dog’ un autre ‘another’ un autre chien ‘another dog’ pomme aussi ‘apple also’ tu fais une pomme aussi ‘you make an apple also’ aussi pomme ‘also apple’

The third occurrence of aussi is found at age 2.4.23, for the first time in an utterance with a verb in right peripheral position: (52) Mother: Anouk: Mother:

qu’est-ce qu’il y a ? ‘what is happening’ met (l)a aussi ‘put it also’ met la aussi, ah le bouchon, mais il est où ton bouchon? ‘put it also, ah the cork, but he is where your cork’

The pattern with aussi in right peripheral position is mostly found in utterances without a verb: (53) celui-là bon aussi ‘that one good also’

An 2;6.11

(54) mama brocoli aussi ‘mummy brocoli also’

An 2;6.11

In the latter example, the subject/topic is affected by the particle, as shown by the following context: (55) Mother:

moi aussi, je mange du broccoli, oui ‘me too I eat broccoli, yes’

At age 2;6.11, for the first time, Anouk produces utterances with aussi not appearing in a peripheral position, but in an utterance-internal position:

JB[v.20020404] Prn:26/06/2003; 10:29

F: SIB2608.tex / p.15 (886-969)

Merging scope particles 

(56) Mother: Anouk:

celui-là il est trop petit oui ‘that one he is too small yes’ celui-là aussi trop petit ‘that one also too small’

Here aussi links two contrastive topic elements (celui-là and celui-là) for which some state of affairs is claimed to hold. Aussi is most frequently found in such an internal position. In utterances without a verb, depending on the context, this internal position could be either analysed as an adjunction to the sentence initial XP to the left of aussi, or as (part of) a FP, linking the two elements on either side (cf. Philippe’s utterance (40)). In the following examples the affected element is underlined: (57) ici aussi Amsterdam ‘here also A’

Anouk 2;7.5

(58) chapeau rouge aussi deux jambes? ‘hat red also two legs’ (59) chez opa et oma aussi bonhomme ‘at granny’s also man’

Anouk 2;7.28

(60) nous aussi chez elle ‘we also at her’

Anouk 2;8.22

There are also a few cases of “doubling”, with aussi both in utterance internal and in final position: (61) moi aussi un gros nez aussi ‘me also a big nose also’ (62) papa met aussi chez opa et oma aussi ‘daddy puts also at granny’s also’

An 2;11.27 An 3;3.17

From age 2;11.13 onwards, Anouk produces utterances containing a finite verb followed by aussi. In those cases aussi clearly occupies a position inside the FP between IP and VP. The affected element (underlined) is either the subject, or an element inside the predicate. Here are some examples: (63) ça c’est aussi une galette ‘that it is also a cake’

An 2;11.13

(64) tu as aussi du beurre dessus ? ‘you have also butter on top’

An 3;3.21

(65) tu te rappelles aussi de Irène ? ‘you remember also Irene’

An 3;6.25

JB[v.20020404] Prn:26/06/2003; 10:29

F: SIB2608.tex / p.16 (969-1032)

 Aafke Hulk

The pattern with aussi in left peripheral position is the least frequent, found mostly in utterances without a verb and with an empty topic; the affected element is either the empty topic (ec) or the other element: (66) ec aussi deux jambes ‘also two legs’ (67) aussi chez papi et mami ‘also at granddad and granny’s

An 2;7.5 An 2;8.22

When we summarize the positions of aussi in all the available files of Anouk, we get the following picture: the internal position is the most frequent (52%) for aussi, the left peripheral (utterance initial) the least frequent (14%). Aussi occurs in utterance final position in 33% of the cases.

. Ook in Anouk’s Dutch As we said above, we have less data in Dutch than in French for Anouk. On the whole we found 41 occurrences of ook. Until age 2;11.13 ook appears slightly more often in utterances without a verb or with only an infinitival verb, than in utterances with a finite verb. There is, however, no clear further development as far as the position of ook is concerned. From the first file, age 2;4.23, onwards ook is found. The first example occurs in right-peripheral position in a mixed utterance without a verb: (68) en ici ook ‘and here also’ In the next occurrences ook occupies an utterance internal position. It is not always clear from the context which is the affected element: In (69) ook seems to function as some kind of affirmative marker, in (70) toetje is the affected element: (69) Anouk ook, Anouk ook eten ‘Anouk also eat’

An 2;5.20

(70) ik ga ook toetje eten ‘I go also sweet eat’

An 2.9.17

We also find ook in left peripheral position, mostly in utterances with the topic missing/empty. Again, it is not always clear which is the affected element – in (72), for example, ook seems to function as a mere affirmative marker: (71) ook oranje ‘also orange’

An 2;10.11

JB[v.20020404] Prn:26/06/2003; 10:29

F: SIB2608.tex / p.17 (1032-1106)

Merging scope particles 

(72) ook met dobbelsteen gooien ‘also with dices throw’ (73) ook in kleuren ‘also in colors’

An 2;10.11 An 3;7.29

In the last file we find an adult like example with ook in right peripheral position, where the object dat has scrambled over it (cf. (24)):18 (74) vind jij dat ook? ‘find you that also’

An 3:10

In earlier files we also find ook in right peripheral position, but these do not involve scrambling of the object. Most of these are two word utterances of the following type, which are also very common in monolingual Dutch children (see above) – ook is adjoined to the affected element: (75) deze ook ‘this one also’

An 2;11.27

However, there are also some cases with right peripheral ook which are very unDutch-like:19 (76) beetje vla mama ook ‘bit pudding mummy also’

An 2;9.17

(77) erg veel kleurtjes ook ‘very many colors also’

An 2;10.11

(78) Mara nu op vakantie ook ‘M now on vacation also’

An 2;11.13

(79) een heel ver land ook ‘a very far away country also’

An 3;6.25

These remind us of their French equivalents with aussi in utterance final position. When we summarize the positions of ook in Anouk’s data, we get the following picture: almost half of all the ook’s (46%) appear in utterance internal position, 23% in left peripheral, utterance initial position and 31% in utterance final position – sometimes in an unDutch-like way.

. Comparison with monolingual children Dutch When we compare the 41 occurrences of ook in Anouk’s files with the 31 occurences of ook in the first 9 files of the Dutch monolingual Laura, we see the following:

JB[v.20020404] Prn:26/06/2003; 10:29

F: SIB2608.tex / p.18 (1106-1181)

 Aafke Hulk

As for the syntax, – –

–

Just as Laura, Anouk produces ook first in the right-peripheral position (XPook), then in the internal position and thirdly in the left-peripheral position. Just as in Laura, in Anouk ook appears most frequently in the internal position, although Anouk has less left peripheral ook’s than Laura, and all the rightperipheral ook’s in Laura’s early utterances without a verb take the form: ik(ke) ook. Anouk only produces one utterance of this form. Four instances of the right peripheral ook in Anouk ((76)–(79)) are of a type not found in Laura, they are “un-Dutch”, more “French-like”.

As for the semantics/pragmatics, a far as we can see in relation to the context, both in Anouk and in Laura, ook functions as a focus-particle and as a topic-particle from early on. In Anouk we also found some examples of ook as an affirmative marker – just as has been mentioned for monolingual Dutch children. However, the numbers are low and the age ranges of Anouk and Laura studied here differ a lot – therefore we cannot make very hard conclusions. More research is necessary.

French When we compare the occurrences of aussi in Anouk, with those in the first 20 files of the French monolingual Philippe, we see the following: As for the syntax, – – – –

aussi appears first in right peripheral position and then in internal position, both in Anouk and in Philippe Anouk has less utterance final aussi’s (33%) than Philippe (66%) Anouk has more and earlier internal aussi’s (52%) than Philippe (31%) Anouk has more left peripheral aussi’s (14%) than Philippe (2%)

As for the semantics/pragmatics, both in Anouk and in Philippe aussi behaves as a topic-particle and as a focusparticle from early on.

Cross linguistic influence? French and Dutch do not differ with respect to the semantics/pragmatics of aussi and ook. Therefore cross-linguistic influence in the sense of using a particle in language A with the semantics/pragmatics of language B does not come into play here. It would be interesting though to examine other, similar scope-taking, discourse

JB[v.20020404] Prn:26/06/2003; 10:29

F: SIB2608.tex / p.19 (1181-1263)

Merging scope particles 

particles, which are also produced frequently in early child language, and which do differ in their semantics/pragmatics, such as e.g. encore and nog. As for the syntax, the two languages clearly develop separately in Anouk. French and Dutch differ with respect to the positions in which aussi and ook can appear, although there is also some overlap. The utterance final position which is (very) frequent for aussi in French, is possible for ook in Dutch only under (very) restricted conditions. Not surprisingly, on the one hand, we found some utterances in Anouk’s Dutch (76)–(79) where she appears to over-use the utterance final position in a non-Dutch-like way, possibly under the influence of French. On the other hand, we found that in French, Anouk uses the utterance final position less than the monolingual Philippe. A more subtle influence may have played a role in the high percentage of utterances with an internal aussi in Anouk’s French. She uses this position more frequently than does the monolingual child Philippe. We have seen above that when such utterances do not contain a finite verb (40), they are structurally ambiguous and moreover they are (superficially)identical to their Dutch (child) equivalents. Consider the following two examples: (80) chez oma aussi bonhomme (81) oma ook sneeuwpop ‘at granny’s also snowman’ In the child-French utterance (80) aussi can either be right-adjoined to the preceding constituent chez oma (82) or be generated in a (precursor of a ) FP (83): (82) [chez oma aussi] [bonhomme] (83) [chez oma] [aussi ] [bonhomme] In the child-Dutch utterance (81) ook can either be left-adjoined to following constituent sneeuwpop (84), or, just as in French, be generated inside a (precursor of a) FP (85): (84) [bij oma] [ook sneeuwpop] (85) [bij oma] [ook ] [sneeuwpop] In a stage of language development when the (bilingual) child is still struggling with the specification and use of functional categories (in each of the languages), constructions with aussi/ook such as (80) and (81) which can be used in both languages come in handy. Therefore, it is not surprising that Anouk uses such constructions a lot in her French, even when her mother does not always like them (cf. (1)).

JB[v.20020404] Prn:26/06/2003; 10:29

F: SIB2608.tex / p.20 (1263-1304)

 Aafke Hulk

. Conclusion In this paper we have taken a closer look at the acquisition and development of two discourse particles, ook and aussi, as witnessed in early child production data. These elements have certain properties which make them particularly interesting: semantically and pragmatically, they affect one of the elements in the utterance and relate it to the preceding discourse in a specific way. In that sense they are “bridging” elements, building a coherent discourse. In the early production data, they are adjacent to the affected element, which may be either the topic or the focus of the utterance. Soon, they also appear in other positions. Syntactically, they have been analysed as functioning as precursors of (verbal) functional categories (which are often lacking in early child speech). We have seen that this may indeed be the case when they appear in the second position of a three word utterance without a verb. This position is used more frequently by monolingual children in Dutch than in French. We know from the literature that the interpretation of (other) “bridging” elements is problematic for children up to a rather advanced age. However, on the one hand, Avrutin and Coopmans (2000) observe that in their experiments four year old children had an adult-like command of bridging. They show that even younger children were sensitive to discourse prominence and they suggest that the younger children’s problems may be related to their inability to implement the available knowledge of discourse rules. On the other hand, it is also well known that certain phenomena which create problems in comprehension (e.g. the possibility of binding and coreference of personal pronouns), appear correctly and earlier in production. It will therefore be very interesting to test the interpretation of utterances containing elements such as aussi and ook in further research, in order to examine whether the dichotomy between production and comprehension also holds for these elements. Finally, we also examined the question whether the use of aussi/ook by a bilingual French/Dutch child was subject to cross-linguistic influence. We argued that the specific language pair lead us to expect this to be possible in the syntactic domain, but not in the semantic-pragmatic domain. Our results indeed showed a higher frequency of aussi in utterance-internal position in the bilingual child Anouk than in the monolingual child Philippe. We argued that this might be explained not only because the internal position is shared by both languages and may therefore be the most economical position for Anouk to use, but also because there might be a more indirect influence of Dutch that could be related to the assumed function of pre-cursor of a functional categorie which holds for both aussi and ook in this position in (monolingual) child speech. It will be interesting to look at these phenomena in other language pairs and at other discourse particles in French/Dutch bilingual children in further research.

JB[v.20020404] Prn:26/06/2003; 10:29

F: SIB2608.tex / p.21 (1304-1361)

Merging scope particles 

Notes * A first version of this paper was presented at the Euroconference on The Dynamics of Learner Varieties in Sant-Feliu (Spain), October 2000. We are very grateful to Henriette Hendricks and Christine Dimroth for commentaries and suggestions on preliminary versions. . In many languages the affected element will receive a pitch accent. We leave the interaction with phonological/phonetic organisational principles outside the scope of this article. . We will not go into the possible LF-movements involved in relation to their status as focus-particle or topic particle (see von Stechow 2000 for an extensive discussion). . According to Perrin-Naffakh (1996) aussi can never appear in sentence intitial position: *aussi Marie pleure (‘also Marie cries’). However Watorek and Perdue (1999) cite sentences of the following type where aussi appears in first position, preceding a spatial PP: aussi dans la rue il y a un autobus (‘also in the street there is a bus’). . However, if we want to maintain the general idea that scope bearing elements have scope to their right (von Stechow), another solution has to be found, e.g. we could assume that a QR-like movement takes place at LF. . This position is not possible when the subject is a clitic: *tu aussi travailles dans le jardin (‘you also work in the garden’). . Here, but not in (6), one could argue that the fronted constituent with aussi is in a position outside the actual sentence (CP), just as in other left dislocation sentences such as: moi aussi j´adore ça (‘me too I adore that’). . Alternatively we could argue that for cases such as (8) and (9), aussi is the head of a functional projection, a ScopePhrase, in the left periphery of the clause. The affected XP would be in the Specifier position of such a ScopeP. . When the finite verb is a copula, followed by an adjectival predicate, this position is less well accepted, according to Perrin-Naffakh (1996): (i)* ? Marie devient aussi paresseuse (‘Marie gets also lazy’). She relates this to the possible confusion with sentences where we find another aussi, the one which expresses a comparative degree: (ii) Marie est aussi paresseuse que toi (‘Marie is as lazy as you’). . In modern French aussi cannot appear in a negative sentence: *Marie n’a pas aussi dormi (‘Marie has not also slept’). In such cases we get non plus: Marie n’a pas non plus dormi (‘Marie has not slept either’). . Left adjunction to VP is another possibility, but it would then be unclear why we find right adjunction in the other cases in French and left adjunction here. . If we want to adopt some kind of “Affect Criterion”, we would have to assume that at some stage of the derivation, the affected XP is in the Specifier position of this Scope Phrase, cf. proposals concerning (the scope of) negation (Haegeman 1995). . In that case, this ScopePhrase itself probably cannot be a functional projection dominating CP, but has to be as a whole in the Specifier position of CP.

JB[v.20020404] Prn:26/06/2003; 10:29

F: SIB2608.tex / p.22 (1361-1446)

 Aafke Hulk . In (23) ook can affect either the subject or the object; (24) however is different – it has the flavour of an echo-question. . The scenario sketched here is very similar to what is proposed, in other terms, in Dimroth et al. (this volume). They analyse elements such as ook as ‘lexical linking devices’, used to validate relations between the predicate and the topic, as precursors in the acquisition of functional categories such as finiteness. . Something similar has been proposed by Hoekstra and Jordens (1994) for the creation and use of the (novel) elements kannie and magnie in child Dutch. Hoekstra and Jordans claim that these modal-like elements are first adjoined to VP and later reanalyzed as heads. . Spontaneous speech production data by the Dutch child Laura available on CHILDES. . This utterance is produced in a context where Philippe and his interlocutor are talking about broken pencils. Là could either be short for celui-là (“that one”) or could indicate a location (“there”). In both cases it expresses a change of topic. . Another interesting example with scrambling is found at age 2;11.13: mag jij deze ook stukjes? ‘may you these also parts’. Similar errors have been reported for monolingual Dutch children (cf. van Kampen op.cit.). . There is no reason to assume that scrambling took place here.

References Avrutin, S. & Coopmans, P. (2000). Children who build bridges. In S. C. Howell, S. A. Fish, & T. Keith-Lucas (Eds.), Proceedings of the 24th annual BUCLD (pp. 80–91). Somerville, MA: Cascadilla Press. Benazzo, S. (this volume). The interaction between verb morphology and temporal adverbs of contrast. Brain, M. D. S. (1976). Children’s First Word Combinations. Monograph for the Society for Research in Child Development, 41(1). Dimroth, C. (2000). Topic and focus particles. Paper presented at the Euroconference on Information Structure, Linguistic Structure and the Dynamics of Acquisition, San Feliu, Spain. Dimroth, C., Gretsch, P., Jordens, P., Perdue, C., & Starren, M. (this volume). Finiteness in Germanic languages: a stage-model for first and second language development. De Houwer, A. (1990). The acquisition of two languages from birth: a case study. Cambridge: Cambridge University Press. Döpke, S. (1998). Competing language structures: the acquisition of verb placement by bilingual German-English children. Journal of Child Language, 25, 555–584. Gayraud, F. (1997). Emergence et développement du placement des particules de portée. Unpubl. paper, Université de Lyon2 & CNRS. Gawlitzek-Maiwald, I. & Tracy, R. (1996). Bilingual bootstrapping. Linguistics, 34, 901–926. Genesee, F. (1989). Early bilingual development: One language or two? Journal of Child Language, 16, 161–179. Haegeman, L. (1995). The Syntax of Negation. Cambridge: Cambridge University Press.

JB[v.20020404] Prn:26/06/2003; 10:29

F: SIB2608.tex / p.23 (1446-1539)

Merging scope particles 

Hoekstra, T. & Jordens, P. (1994). From Adjunct to Head. In T. Hoekstra & B. Schwartz (Eds.), Language Acquisition Studies in Generative Grammar (pp. 119–150). Amsterdam and Philadelphia: John Benjamins. Hulk, A. C. J. (1997). The acquisition of French object pronouns by a Dutch/French bilingual child. In A. Sorace et al. (Eds.), Proceedings of GALA (pp. 521–527). Edinburgh. Hulk, A. C. J. (2000). Non-selective access and activation in child bilingualism: the syntax. In S. Döpke (Ed.), Cross-linguistic structures in simultaneous bilingualism (pp. 57–78). Amsterdam and Philadelphia: John Benjamins. Hulk, A. C. J. & Müller, N. (2000). Bilingual first language acquisition at the Interface between Syntax and Pragmatics. Bilingualism: Language and Cognition, 3(3), 227–244. Jordens, P. (to appear). Finiteness in early Dutch. Linguistics. Kampen, J. van (1997). First Steps in Wh-Movement. PhD, Utrecht University. Karmiloff-Smith, A. (1981). The grammatical marking of thematic structure in the development of language production. In W. Deuch (Ed.), The child’s construction of language. New York: Academic Press. MacWhinney, B. & Snow, C. (1985). The Child language Data Exchange System. Journal of Child Language, 12, 271–296. Meisel, J. (1989). Early differentiation of languages in bilingual children. In K. Hyltenstam & L. Obler (Eds.), Bilingualism across the lifespan: Aspects of acquisition, maturity and loss (pp. 13–40). Cambridge: Cambridge University Press. Müller, N. (1998). Transfer in bilingual first language acquisition. Bilingualism: Language and Cognition, 1(3), 151–171. Müller, N., Hulk, A. C. J., & Jacubowicz, C. (1999). Object Omissions in Bilingual Children: Evidence for Crosslinguistic influence. In A. Greenhill, H. Littlerfield, & C. Tano (Eds.), Proceedings of the 23rd BUC (pp. 482–494). Somerville, MA: Cascadilla Press. Müller, N. & Hulk, A. C. J. (2001). Crosslinguistic influence in bilingual language acquisition: Italian and French as recipient language. Bilingualism: Language and Cognition, 4(1), 1–23. Nolke, H. (1993). Le regard du locuteur. Paris: Klimé. Paradis, J. & Genesee, F. (1997). Syntactic acquisition in bilingual children: autonomous or interdependent? Studies in Second Language Acquisition, 18, 1–15. Penner, Z., Tracy, R., & Weissenborn, J. (1995). Where scrambling begins: triggering object scrambling at the early stage in German and Bernese Swiss. Unpubl. paper. Perrin-Naffakh, A.-M. (1996). Aussi adjonctif: de la syntaxe à la sémantique. Le français moderne, LXIV(2), 136–159. Powers, S. (1996). MAPping Phrase Markers. In C. Koster & F. Wijnen (Eds.), Proceedings of GALA. Groningen. Roeper, T. (1995). The Role of Merger Theory and Formal Features in Acquisition. In H. Clahsen (Ed.), Generative Approaches in Language Acquisition (pp. 415–451). Amsterdam and Philadelphia: Benjamins. Stechow, A. von (2000). Topic particles? Unpubl. paper. Tracy, R. (1995). Child Languages in Contact: The simultaneous acquisition if two first languages (English/German) in early childhood. Habilitationsschrift Universitat Tübingen.

JB[v.20020404] Prn:26/06/2003; 10:29

F: SIB2608.tex / p.24 (1539-1544)

 Aafke Hulk

Watorek, M. & Perdue, C. (1999). Additive particles and focus: observations from learner and native-speaker production. Linguistics, 37(2), 297–323.

Creole prototypes as basic varieties and inflectional morphology Angelika Becker and Tonjes Veenstra

.

Introduction

There has long been a bi-directional symbiotic exchange between the study of creole languages and the study of second language acquisition with each subdiscipline contributing concepts as well as results to the other (cf. Véronique 1994 and Muysken 2001 for a recent overview). In this paper we bring together two recent lines of research from each discipline to gain some new insights with respect to the grammatical properties of creole languages as well as the agents and processes involved in the genesis of these languages. One of the questions that we would like to address here is to what extent the grammatical properties of creole languages can plausibly be attributed to what we know of the processes of L2 acquisition of their lexifier languages. If one assumes traditional classifications of morphological systems in which languages are judged on the basis of their inflectional subsystems rather than on the full range of their morphological processes (Spencer 1991), then creole languages from the Caribbean tend to be classified as isolating languages. They have not only little or no verbal inflection, i.e. categories as tense and aspect tend to be expressed by periphrasis, but also little or no nominal inflection, i.e. case tends to be distinguished by word order, gender concord is in general not marked, and categories like plurality and diminutive are marked by periphrasis.1 Basically the same observations have also been made for the results of untutored second language acquisition, i.e. early “interlanguages” (Selinker 1972; Perdue 1993). These are relatively stable and well-structured forms of language without functional inflection of words (inflectional morphology only starts to find its way into these varieties at subsequent stages of development, if at all). Because of this observed parallelism, we limit the discussion to inflectional morphology. The first line of research from the study of creole languages is the proposal by McWhorter (1998) that there is a creole prototype. Although Muysken (1988)

 Angelika Becker and Tonjes Veenstra

already argued against the existence of a creole type resulting from a language bioprogram (as implied by the work of Bickerton 1981, 1984) by pointing out a significant amount of variation in syntactic structures found in the Caribbean creoles (which all would count as proto-typical in McWhorter’s sense), McWhorter claims that most creoles exhibit three (morphophonological) traits, whose unique combination is not found in any other language. Therefore, he argues that it is possible to regard creoles as a unique typological class, to be structurally distinguished from non-creole languages. The three features are: (i) little or no inflectional morphology, (ii) little or no use of tone to lexically contrast or encode syntax and (iii) semantically regular derivational affixation. This proposal has met with severe criticism (cf. DeGraff 2001 and Muysken & Law 2001). Although there is no doubt that many creoles have (very) little inflectional morphology, it is doubtful whether this is due to a creole prototype. First, the superstrate and substrate languages of the Atlantic creoles are (or were) themselves not very rich in inflection either. Second, as Muysken and Law (2001: 49) note, “[i]f we take creolization in typologically very different languages such as Shaba Swahili (de Rooy 1995, 1996) and Amazonian Ecuadorian Quechua (Muysken 2000) there is simpliflication and regularization of inflection, but not loss of inflection.” This latter observation we will corroborate in this paper, by showing that even in what would count for McWhorter as prototypical creoles (Haitian, Morisyen, etc.) one does not find an across-the-board loss of inflection. Thus, it will be clear that McWhorter’s first claim cannot be upheld in its strong form (i.e. no inflectional morphology will survive the creolization process). The question why this type of morphology gets marginalized and minimalized in creoles, remains, however. Therefore, we will focus on the survival/retention and reappearance of inflectional morphology in French-related creoles. The second line of research from the study of second language acquisition is the proposal by Klein and Perdue (1992, 1997) that there is a Basic Variety. The concept of the Basic Variety was developed in the course of an extensive cross-linguistic longitudinal project on untutored Second Language Acquisition.2 The forty learners included in the study represented ten different source language-target language pairs.3 During the thirty-month period of observation it became apparent that all the learners developed a specific language system – a basic variety – which is simple, well-structured, and communicatively very effective. The Basic Variety has a high potential of fossilization: for about a third of the learners, the linguistic development stopped at this level. The structural properties of this linguistic system are by and large independent of the target as well as the source languages. Characteristic features of the system include: (i) a lexical repertoire consisting mainly of lexical categories (open class items) and only a small list of function words, (ii) an utterance organisation determined by the interaction of a small number of syntactic, semantic and pragmatic constraints and (iii) there is hardly any (if at all) functional inflection of words. This latter property it shares with the “creole prototype”.

Creole prototypes as basic varieties and inflectional morphology 

Having set the stage for the discussion, we would like to add a final note on the definition of inflectional morphology that we adopt throughout the rest of the paper. In most definitions of inflectional morphology grammatical notions like tense, aspect, agreement, case, etc. are included in the definition (cf. Scalise 1984; Spencer 1991; Radford et al. 1999; Fromkin 2000). We opt for a definition in which these notions do not play any role, and accordingly define it as follows: Inflectional morphology is a system of oppositions between non-category-changing morphemes with (grammatical) meaning whose presence is mandated by specific sentence structures. In this paper we focus on the remnants of verbal inflectional morphology that we find in French-related creoles. Specifically, we will deal with two questions: (i) To what extent can the properties of verbal morphology in French-related creoles be attributed to the process of second language acquisition (ii) what conclusions can be drawn for the process of creole genesis. Although all current theories of creolization regard second language acquisition as one of the determining factors in creole genesis, the assumptions about its impact and the way it interacts with other contact phenomena like transfer, relexification, language shift, language (re-)creation, etc. differ widely (DeGraff 1999; Mufwene 1990; Muysken 1981, 2001; Siegel 1999; Thomason 2001; Véronique 1994). We will argue that “creolization” requires a three-generational scenario of language shift4 (Bickerton 1977; Corne 1999; Roberts 2000), thereby bringing it more into line with other (albeit less dramatic and drastic) language shift situations (Macías 2000; Veltman 2000). The difference between creolization and these other language shift situations is in our opinion due to a target shift in the case of creolization (see Baker 1990, 1996 for extensive argumentation of this point): due to demographic and sociological factors at a certain point in time in the history of a colony (Baker & Corne 1986), the African slave population did not stick to the European superstrate languages as the target language, but turned to the emerging contact variety instead. We also argue that “creolization” is a very complex phenomenon not to be accounted by one general mechanism, but consisting of at least two different processes, each with its own agent (cf. Veenstra 1996: 195). On the one hand, with the first generation we have untutored second language acquisition, in which the adults (i.e. the different cohorts of slaves imported from West Africa) are the agents. This process is responsible for some of the basic-variety effects in the grammars of creole languages. On the other hand, with the second generation there is bilingual first language acquisition, with the children (i.e. the second generation, consisting of the “creole” population born in the colony) being the agents. In this process we find all kinds of relexificational effects (Lefebvre 1998; Lumsden 1994, 1995, 1999a, b).5 The third generation is the first generation of monolingual speakers of the creole. Such a scenario is highly compatible with the observations by Muysken (1996) that creole languages in initial stages exhibit less “Africanisms” than in later

 Angelika Becker and Tonjes Veenstra

ones. Such diachronic developments, on the other hand, are hard to accomodate in more traditional scenarios of creole genesis, in which there is only one process of creolization. The paper is organized as follows. After a concise description of the inflectional system of French, we show in Section 1 how the acquisition of inflectional morphology takes place in learner varieties of French from a Basic Variety-perspective. In Section 2 we discuss the reflexes of French inflectional morphology that survived the creolization process and show up in a number of French-related creoles. These reflexes take the form of a long/short opposition in verb forms. We show that the different creoles have chosen different grammatical properties to be marked by this opposition. In addition, we compare Morisyen and the Portuguese-related creole of São Tomé and argue that both creoles mark the same grammatical property, namely an argument/adjunct distinction, and show that in both cases it involves a (morphophonological) process that takes places after the syntax, and is independent of the superstrate as well as substrate languages that were present in the language contact situation. In Section 3 we bring together the findings of the previous two sections, discuss the implications they have for theories of creole genesis, and provide further evidence for the three-generational scenario of the creolization process with multiple agents proposed above.

. Inflectional morphology in French and its basic varieties The following presentation of the acquisition of verbal morphology in French learner varieties is based on the findings of the project Second Language Acquisition of Adult Immigrants (see Note 2). The results relevant to the present discussion are published in Klein and Perdue (1992, 1997), Dietrich, Klein, and Noyau (1995), and Starren (2001). The study included learners of French with Moroccan Arabic and Spanish as source languages. In untutored second language acquisition learners have to segment and analyse the lexical and functional information encoded on the verb on the basis of mainly oral input. The learners’ task of analysis is especially difficult in the acquisition of French for two basic reasons. First, there are a number of homophonous forms which carry different values (such as tense, aspect, mood, person, number). Second, relevant functional information is often encoded in a preverbal auxiliaryclitic pronoun-cluster which is in most cases perceived as a single prosodic unit (cf. Dietrich, Klein, & Noyau 1995). We briefly discuss these two problem areas. The distinction of morphological variants of a verb has been retained in the orthographic representation of French but the number of audibly distinctive forms in colloquial French is severely reduced (cf. Harris 1987). This can be illustrated by

Creole prototypes as basic varieties and inflectional morphology 

the present indicative forms of the class of -er verbs, exemplified by the verb parler ‘to speak’. The majority of French verbs belongs to this conjugation class: (1) 1sg 2sg 3sg 1pl

je parle tu parles il/elle parle on parle (nous parlons 2pl vous parlez 3pl ils/elles parlent

/parl/ /parl/ /parl/ /parl/ /parl«f/) /parle/ /parl/

There is no audible contrast between the singular forms and the 3pl form. Furthermore, the unambiguous 1pl marking -ons /«f/ is rare in casual speech as the function of the subject pronoun nous ‘we’ has largely been taken over by the originally impersonal pronoun on ‘[lit.] one’. As a consequence only the 2pl form (in our example /parle/) is distinctive. The same holds by and large for the other verbal paradigms of French (cf. the overview in Harris 1987). As can be read from the paradigm in (1), the verb appears either in the root form or with a final /-e/. The latter form (e.g. /parle/) has two more functions: First, it serves as an infinitive which appears, e.g., in the frequently used periphrastic verb constructions like je viens de parler ‘[lit.] I come from speaking’, ‘I just spoke’ or je vais parler ‘I am going to speak’. Second, it carries the value of a past participle which is required in French in a broad range of compound verb constructions like the passé composé (e.g. j’ai parlé ‘[lit.] I have spoken, I spoke’). Thus, a form like /parle/ can have the following functions: (2) /parle/ vous parlez 2pl present indicative /parle/ parlé past participle /parle/ parler infinitive The ending /-e/ contrasts with /-7/ which marks the imparfait for all persons but 2pl (and 1pl if nous is preferred to on), as shown in (3):6 (3) 1sg 2sg 3sg 1pl

je parlais tu parlais il/elle parlait on parlait nous parlions 2pl vous parliez 3pl ils/elles parlaient

/parl7/ /parl7/ /parl7/ /parl7/ /parlj«f/ /parlje/ /parl7/

However, the opposition /-e/ vs. /-7/ is neutralized in many (regional) varieties of French and the imperfect forms are, thus, again homophonous with the forms

 Angelika Becker and Tonjes Veenstra

in (2) (cf. Noyau et al. 1995). It goes without saying that the multifunctionality of a single form confronts the learner with a serious problem in form-functionassignment. The second major problem which learners of French encounter in the analysis of the verb complex concerns the unstressed clitic pronouns which immediately precede the finite verb component (in declaratives). Use of the clitic subject pronoun which tends to be phonetically reduced in casual speech (e.g. /il/ > /i/ (3sg/pl masc), /ty/ > /t/ (2sg)) is obligatory. The rich clitic system of French marks person, number, case and gender of the actants of the verb (Klein & Perdue 1992; Harris 1987). The clitic pronouns not only have deictic and anaphoric functions, they are also required as “pronoun copies”, i.e. they indicate co-referentiality with a right- or left-dislocated verb argument. Dislocation of verb actants is widespread in spoken French. Therefore, the verb frequently appears with a cluster of unstressed, often reduced proclitic pronouns which poses a serious problem for segmentation. This is even more so in the case of auxiliary-lexical verb constructions. The pronouns cliticize on the finite auxiliary, and the whole verb complex (subject pronoun – object pronoun and/or oblique pronouns – auxiliary – lexical verb) is produced as one (phonological) unit (e.g. je te l’ai dit /Štledi/ ‘[lit.] I you it have told’). Noyau et al. (1995: 147) summarize the two essential problems of analysis learners are confronted with in the following way: “(i) the suffix area is opaque (. . . ); (ii) the prefixed area includes a cluster of entangled unstressed markers (. . . )”. In the remainder of this section we discuss how the learners of the ESF project cope with these problems. In the earliest stages of acquisition the majority of the utterances are verbless. The utterances of beginners “are extremely simple and mainly consist of seemingly unconnected nouns, adverbs and particles (. . . )” (Klein & Perdue 1992: 302). All the learners studied go beyond this level and develop the learner system which interests us in the present context: the Basic Variety. At this stage of development, utterances are characterized by the presence of verbs. The verb forms, however, do not show any functional variation. Formal variation can occur, but it has no functional value, i.e. there is no marking of person, number, tense or aspect by inflectional morphology. In the basic variety of both the Moroccan and the Spanish learners, verbs mainly appear in two formal variants with respect to their endings: as a short form V-/ø/ or as a long form V-/e/.7 Some verbs show up in both forms but the variation does not have any functional value. This can be illustrated with examples from the data of the Moroccan learner Zahra (see the detailed discussion of Zahra’s development in Noyau et al. (1995: 157ff.), and in Klein and Perdue (1992: 237ff.):8

Creole prototypes as basic varieties and inflectional morphology 

(4) /-rogard/ vs. /-rogarde/ (Fr. regarder ‘to look’) a. [context: retelling of a Charlie Chaplin movie; a lady observes a girl who is stealing a loaf of bread] et après la-dame l’autre /le-rogard/ la-dame le-volur (. . . ) and after the lady, the other one, look at the lady, the thief (Klein & Perdue 1992: 243) b. [context: retelling of a Charlie Chaplin movie; Charlie is looking at a lady] et après charlot /i-rogarde/ la-dame and then charly look at the lady (Klein & Perdue 1992: 242) (5) /-werw/ vs. /-werwe/ (Fr. chercher ‘to look for’) a. [context: retelling of a Charlie Chaplin movie; the girl who has stolen the loaf of bread runs away followed by a policeman] la police /i-werw/ à-la-fille (. . . ) the police look for the girl (Klein & Perdue 1992: 247) b. [context: retelling of a Charlie Chaplin movie; the girl is looking for a house in which she can live together with Charlie] et après (. . . ) la-dame /ele-werwe/ à-la-maison and then the lady look for the house (Klein & Perdue 1992: 238) (6) /-part/ vs. /-parti/ (Fr. partir ‘to leave’) a. [context: retelling of a Charlie Chaplin movie; Charlie and the girl could escape from the police; they run away and come to an idyllic middle-class residential area] tous les deux après /i-part/ à-la-campagne both of them then depart to the countryside (Klein & Perdue 1992: 247) b. [context: same as in (6a): (6b) is a reformulation of (6a)] /le-parti/ tous les deux à-la-campagne depart both of them to the countryside (Klein & Perdue 1992: 247) As the examples show, the formal variation is not motivated by person or number marking. The examples in (4) through (5) all have a third person singular subject, but the examples in (6) have a third person plural subject. It can also be excluded that the two forms serve to express a temporal contrast. Thus, in the following examples, Zahra always uses a (variant of the) short form of Fr. rester ‘to stay’, although she talks about states/events which are situated in the present (7a), the future (7b) and the past (7c): (7) a.

[context: personal narrative; the husband of a friend wants to obtain a divorce, because he wants to marry another woman living in Mo-

 Angelika Becker and Tonjes Veenstra

rocco; his wife, Zarah’s friend, is living in France] et maintenant /rest/ la-france la-femme and now live the France the wife (Noyau et al. 1995: 162) b. [context: same as above; the husband will leave for Morocco and live there with the new wife] /ile-parti/ le maroc et après /res/ la-dame du maroc depart Morocco and then stay the lady from Morocco (Noyau et al. 1995: 162) c. [context: personal narrative; Zahra talks about her migration to France; her husband went first while she was still staying in Morocco] pendant ce-temps-là tu étais où? target language speaker during that time where were you? le maroc; /res/ le maroc avec les enfants zahra Morocco; stay Morocco with the children (Noyau et al. 1995: 163/4) The examples illustrate a general trait of the Basic Variety: temporal relations are not expressed by means of inflectional morphology (cf. Dietrich et al. 1995; Klein & Perdue 1997).9 In contrast to the verbs used by the Spanish learners, the verbs of the Moroccan learners show a variety of verbal prefixes (see (4)–(6) above). The genetic relatedness of French and Spanish obviously gives the Spanish learners a head start in the identification of the lexical verb component. This is not the case for the Moroccan learners, and, as a result, the verbs appear with a broad range of prefixes, eg. /i/, /li/, /le/, /ili/, /e/ etc. Together with the short/long variation this results in a considerable number of formal variants. In a narrative of one learner, for example, the French verb dormir ‘to sleep’ shows up in the following forms: /dorm/, /e-dorm/, /e-dormi/, /ila-dormi/, /la-dorm/, /li-dorm/, /li-dormi/ (cf. Noyau et al. 1995: 171). In all these instances the subject is 1sg and the narrated situation is in the past. It seems reasonable to assume that the prefixation is a reflex of the French clitic pronoun-auxiliary system, but in the learners’ Basic Variety a systematic marking of gender, number, person or tense is not established. The learners take up forms (or one form) that occur frequently in the input and use them as unanalyzed chunks. The formal variation attests the learners’ “awareness” that formal distinctions exist, but “there is no early, uniform tendency to form functional distinctions with them” (Noyau et al. 1995: 205). With both groups of learners development beyond the Basic Variety is slow. During the thirty-month period of observation, the establishment of morphological distinctions approaching those of the target language (French) was achieved by just a few (fast) learners, but only towards the end of the data collection period. Especially the Moroccan learners tested a range of hypotheses about the possible functional (aktionsart, aspectual, or tense) contrasts between V-/ø/ and V-/e/ forms

Creole prototypes as basic varieties and inflectional morphology 

but the solutions found are short-lived and do not conform to target language oppositions (cf. Noyau et al. 1995).10 According to Starren (2001), learners do not approach the target language system of tense and aspect marking by developing verbal morphology that marks the relevant distinctions. Rather, the development towards the target language system proceeds via the introduction of analytical constructions which consist of a free lexical element plus the base form of a lexical verb, for example: /komãs/ + V (derived from Fr. commencer à ‘to start to’), /sefini/ + V (derived from Fr. c’est fini ‘it’s over’) or /ve/ + V (derived from Fr. (je) vais + infinitive ‘I am going to. . . ’). The first component of the construction which Starren (2001) calls a “proto-auxiliary” conveys aspectual or tense distinctions.11 The pattern paves the way for target-language-adequate auxiliary-verb constructions which are opposed to simple verb forms. This allows the marking of past (auxiliary avoir/être + V) vs. present (bare V). In both these cases, however, the lexical verb still shows variation between a short and a long form. Person and number distinctions appear on the auxiliaries avoir/être ‘have/be’, but a systematic differentiation between finite and infinite forms of lexical verbs is only incipient. Summarizing, we can say that although at the stage of second language acquisition identified as the Basic Variety the learners’ output already exhibit formal variation in the verb forms, they do not seem to have attached the functions as found in the target language to it. As we will show in the next section, this variation in form (basically ø vs. -e) has survived in some of the French-related creoles, although with different functions.

. The long/short opposition in French-related Creoles The rich inflectional paradigm of the different verb classes in French hardly left any trace in the French-related creoles. Some of them exhibit the long/short opposition (ø vs. -e), but only in a certain set of verbs (see below). Although this might be traceable in part to regional dialects of French in which, as opposed to the European standard language, the paradigms are (and were) much eroded to the extent that, for instance, the present tense paradigm of verbs of the first conjugation class (i.e. verbs ending in -er) is basically reduced to the opposition ø vs. -e (see above), it is not the case that the creators of the creoles overgeneralized the second person plural ending (or any other person ending for that matter), nor the finite/non-finite opposition. It should be noted that not all French-related creoles exhibit this formal opposition to the same extent, however. Some apparently do not have it at all. Thus, Corne (1999: 132) notes that, with a few exceptions, verb forms in the Lesser Antillean Creoles (the different varieties as spoken in Guadeloupe, Martinique, St. Lu-

 Angelika Becker and Tonjes Veenstra

cia, Dominica, a.o.) are invariable, and what little variation there is has no semantic correlates. For some other French-related creoles it is difficult to get a clear picture of the distribution of this opposition from the existing literature. Thus, Corne (1999) notes that in the case of Reunionais the long/short opposition belies contextual, grammatical as well as (to a large extent) sociological variation. It appears to occur in two guises: (i) in some lects, it is a present/past distinction; (ii) there is a contextsensitive rule that deletes the -e when the following element is governed by the verb. The first pattern seems reminiscent of the pattern found in Louisianais and the latter one seems reminiscent of the Morisyen pattern, both to be discussed below in more detail. DeGraff (2001) contains a short discussion of the pattern in Haitian, showing that in the case the object of a verb is questioned (i.e. has undergone wh-movement and occurs in sentence-initial position) and the verb is the last overt element in the VP, this verb is strongly preferred to surface in the long form (as opposed to when the verb is followed by its complement), a pattern (again) reminiscent (but not identical) to that of Morisyen. Additional factors that complicate the picture in Haitian are the status of the object, i.e. full NPs, but not pronominal objects, induce the short form when following the verb, and emphasis which disfavors the occurrence of the short form. In addition, there appears to be much variation among the different dialects of Haitian. Although the patterns of both Reunionais and Haitian partly look reminiscent of the Morisyen pattern (as discussed in Section 2.2 below), a crucial difference seems to be that in Morisyen the distribution of the long and short forms is categorial and not subject to dialect variation. Due to the relative scarcity of relevant information for these creole varieties we will not discuss them here any further, but instead focus on those varieties (Louisianais, Morisyen) for which we do have the relevant data.

. Louisianais Verbs in Louisianais fall into two classes: I and II (cf. Neumann 1985; Rottet 1992), of which class I verbs form the numerical majority. It is also this class that shows the long/short opposition with interpretative and syntactic correlates. The other group (i.e. class II verbs) does not exhibit this alternation.12 In the (more) basilectal varieties of Louisianais only the long forms are used (on a par with the Frenchrelated creoles of the Caribbean), while the short forms correspond to the Frenchlike forms found in more acrolectal varieties.13 The difference between the long and short forms lies mostly in the truncation of the final -e(r ) in class I verbs, e.g. frem/freme ‘to shut’, kup/kupe ‘to cut’, kuv/kuver ‘to cover’. Neumann (1985) and Rottet (1992) have shown that long and

Creole prototypes as basic varieties and inflectional morphology 

short forms have different temporal/aspectual (TMA) interpretations. The short form occurs in the (habitual) present, the second person imperative and in the complement of fo (a predicate derived from il faut que ‘it is necessary that’). This interpretative difference is illustrated by the imperfective/perfective contrast in the following example (taken from Neumann 1985: 196): (8) a.

Sop-la frem a sez-er. Shop-det close at six-o’clock ‘The shop (always) closes at six o’clock.’ b. Sop-la freme a sez-er. Shop-det close at six-o’clock ‘The shop closed at six o’clock.’

In addition to this interpretative difference, the long/short opposition also correlates with a syntactic difference. On the basis of the standard syntactic tests of Pollock (1989) with respect to adverb and negation placement, Rottet (1992) argues that verbs with the long form remain inside the VP, and verbs with the short form raise out of the VP into the functional domain dominating the VP (a case of V-to-I movement). The relevant data to establish this link between the long/short opposition and verb movement is given below: (9) a.

Mo pa monzhe/*monzh. 1sg neg eat ‘I did not eat.’ OR ‘I have not eaten.’ b. Mo monzh/*monzhe pa. 1sg eat neg ‘I don’t eat.’

Fo tuzhu kupe/*kup zer b-la. FO always cut grass-det ‘It is always necessary to cut the grass.’ b. Fo to kup/*kupe tuzhu zer b-la. FO 2sg cut always grass-det ‘You always have to cut the grass.’

(10) a.

Negation Placement

(Rottet 1992: 268) Adverb Placement

(Rottet 1992: 266)

Rottet (1992) argues that the short forms are inflected for tense. In (8)–(10) the short form is interpreted as (habitual) present and the long one as past tense. The analysis is that the head of a Tense phrase (TP) contains a feature (or an affix) with the relevant temporal interpretation (present), which triggers the V-to-I movement. In the examples with a past tense interpretation (and the long form), on the other hand, the relevant feature in T0 does not induce this movement and the verb remains inside the VP. This latter pattern is akin to the pattern of verb placement

 Angelika Becker and Tonjes Veenstra

as well as to the temporal interpretation of this pattern in Haitian, as extensively discussed by DeGraff (1997). An additional argument for the correlation between the long/short opposition and V-to-I movement is the incompatability of short forms and preverbal TMA markers. If these markers (arguably having a non-affixal status (cf. Veenstra 1996 for Saramaccan)) are present, only the long form can surface: (11) Le klosh ape sone/*son aster . det bell asp ring now ‘The bells are ringing now.’

TMA marker (Rottet 1992: 267)

The short form of the verb requires an (abstract) affixal head in its extended functional domain (in order to check the inflectional features by raising into that domain). If a non-affixal TMA marker is present in this domain, it blocks the raising of the verb. Verbs that remain in-situ (i.e. inside the VP) are spelled out in the long form. The conclusion of this section is that there exists a correlation between the long/short opposition and the intepretation of Tense in Louisianais. Syntactically, this correlation shows up as a verb movement rule (short forms are only licensed if the verb is moved to INFL).

. Morisyen Although Corne (1999: 196) notes that the verb form in the Isle de France creoles in general and in Morisyen in particular is basically invariable, there is a very large class of verbs, ending in -i, -o, and -e (of which the latter is the clear majority) that have a long form with the final vowel and a short one without it.14 As such, the situation on the formal side of the alternation in Morisyen is basically the same as in Louisianais. With respect to the grammatical property this alternation marks, the two creoles differ, however. In Morisyen there is no correlation with the temporal/aspectual interpretation of the sentence nor does the alternation correlate with different word order patterns as discussed in Section 3.1. On the other hand, it seems to involve grammatical funtion-changing operations (e.g. active/passive alternation) and argument-selection (argument/adjunct distinction). We discuss them in turn.

.. Active/passive alternation Corne (1981) notes that one of the contexts in which the long/short opposition shows up is the active/passive alternation, as shown in example (12):15

Creole prototypes as basic varieties and inflectional morphology 

(12) a.

Zot burzua in fek pey zot. 3pl boss asp just pay 3pl ‘Their boss has just paid them.’ b. Zot in fek peye par zot burzua. 3pl asp just pay by 3pl boss ‘They have just been paid by their boss.’

active

passive

One might argue that the “displaced” position of zot ‘3pl’ in (12b) mirrors that of Jimi in the passive sentence Jimi was murdered, and so does the morphological contrast between pey in (12a) and peye in (12b) mirror that between the active murder and passive murdered in English. If one assumes that this alternation is derived in the syntax (cf. Baker 1988; Baker, Johnson, & Roberts 1989), it would follow that the long/short opposition marks the grammatical-function change, i.e. it is a morphological marker of (syntactic) passivization. Corne (1981), however, takes a different track, and argues that in case the long form shows up, the element is not a verb but an adjective, since verbs, but not adjectives, undergo the syncopation rule (i.e. exhibit the long/short opposition). In other words, he assumes a lexical analysis of passivization, and claims the long/short opposition to be due to a difference in categorial status. Thus, the issue boils down to which analysis fits best the available data on active/passive alternations in Morisyen. Languages can have verbal passives, adjectival, or both. For instance, English has them both (Wasow 1977): (13) a. The letter was stolen (by John). b. The pillow remained stuffed (*by John).

verbal adjectival

Verbal passives differ from adjectival ones in (at least) three respects (cf. Levin & Rappoport 1986; Verrips 1996): (i) adjectival passives generally cannot have byphrases, verbal ones can, (ii) verbal passives have an implicit argument (i.e. the external argument is syntactically represented in the passive), as opposed to adjectival ones (i.e. the external argument is not represented in the syntax), (iii) adjectival passives generally refer to a state, verbal ones refer to a dynamic event. Note that the first two differences are interrelated. Following proposals of Wasow (1977) and Borer (1984), the basic proposal in the literature to account for these different thematic properties has been (and still is) that adjectival passives are formed in the lexicon, and verbal passives are formed in the syntax. Lexical derivations may involve argument-deletion, hence no implicit argument present in adjectival passives. In syntactic derivations, on the other hand, this is ruled out (due the Projection Principle, or its successor), hence an implicit argument present in verbal passives.

 Angelika Becker and Tonjes Veenstra

Unfortunately, there is no data available on implicit argument effects in Morisyen.16 Applying the by-phrase test, leads to the following result. By-phrases are possible as shown in (12b) above, repeated here for convenience: (12) b. Zot in fek peye par zot burzua. 3pl asp just pay by 3pl boss ‘They have just been paid by their boss.’ Prima facie, this constitutes evidence for the verbal status of the passive. The issue is not as crystal-clear, however. As pointed out in Roeper (1987), the presence versus absence of a by-phrase is not a “hard” property distinguishing verbal and adjectival passives in English (see Cornelis & Verhagen 1995 for the same observation for Dutch). (14) is an English example of an adjectival passive taking a by-phrase: (14) The code remained unbroken by the Russians. Only adjectival passives can undergo un-prefixation and be embedded under verbs like remain. So far there is not enough evidence available at the moment to decide which analysis (lexical vs. syntactic) fits the pattern in Morisyen best, and, therefore, we leave this issue for future research. Independent of the above, however, it is clear that even if it turns out that a syntactic analysis is the most feasible analysis, and the long/short opposition could be argued to instantiate a case of passive morphology, it is not the case that this voice-marking is the sole function of the opposition in Morisyen. First, in addition to passive/active pairs we also find different pairings, such as passive and intransitive/middle/anti-causative pairs that are marked by this opposition.17 Second, the opposition also marks whether the element following the verb is an argument of it or not. We discuss this property in the next subsection.

.. Argument/adjunct distinction The broad generalization that emerges from the studies of Corne (1981), Seuren (1990), Baker and Syea (1991), and Syea (1992) on the long/short opposition in Morisyen, is that verb syncopation applies whenever the verb in question is followed by material of its own VP. In other words, if a verb is followed by a element that it selects, it will appear in the short form. Elements that induce this rule of syncopation include direct objects, indirect objects, measure phrases, selected adverbials, etc. Thus, a verb like manze ‘eat’ selects for an (edible) object, but not for a location, hence if it is followed by an object, it appears in the short form (15a), but if it is followed by a locative PP (e.g. the name of a restaurant, as in (15b)), the long form surfaces (data taken from Baker & Syea 1991):

Creole prototypes as basic varieties and inflectional morphology 

(15) a.

Pye ti manz/*manze min. Peter tns eat Chinese.noodles ‘Peter ate Chinese noodles.’ b. Pyer ti manze/*manz Rozil. Peter tns eat Rose-Hill ‘Peter ate at Rose-Hill.’

If, on the other hand, the locative follows a verb of movement, the verb has the short form: (16) Pyer ti al/*ale Rozil. Peter tns go Rose-Hill ‘Peter went to Rose-Hill.’ It may be the case that some (or most) of the data discussed in Section 2.2.1 can be subsumed under the same generalization. For instance, if a verb is followed by its object (as in active sentences) it has the short form, but when the object appears in the subject position (as in passive sentences), it has the long form. If so, this would mean that nothing particular hinges on the choice between a lexical and a syntactic analysis of passivization. The application of this syncopation-rule is post-syntactic. If the object of a verb is wh-moved to the left periphery of the sentence, the verb shows up in the short form, as shown in (17): (17) Q Ki Pye ti manze/*manz? A Pye ti manz/*manze min. What Peter tns eat Peter tns eat Chinese noodles ‘What did Peter eat?’ ‘Peter ate Chinese noodles.’ The conclusion we draw from the discussion in this section so far, is that the long/short opposition in Morisyen marks the status of the element following the verb, i.e. whether it is selected by the verb or not (argument vs. adjunct). Thus, the selectional properties of the verb are morphologically marked. Interestingly, Morisyen is not the only creole to mark the selectional properties of its verbs. In the Portuguese-related creole São-Tomense we find the same pattern. Thus, a verb like ‘to go’ has two forms, ba and be. These two forms are in complementary distribution (like the long and short forms in Morisyen). If the verb ‘to go’ is followed by a locative element it surfaces as ba (18a), but when it is followed by a comitative (arguably a non-selected element) it surfaces as be (18b). The example in (18c) shows that the categorial status of the following element (noun vs. preposition) does not play a role. All the data are taken from Hagemeijer (2000):

 Angelika Becker and Tonjes Veenstra

(18) a.

E ba/*be ke. 3sg go home ‘He went home.’ b. E be/*ba ku bo. 3sg go with 2sg ‘He went with you.’ c. E ba/*be nglentu ke 3sg go inside house ‘He went into the house.’

Although Hagemeijer (2000) claims this suppletion of forms depends on the telicity of the predicate (telic = ba, atelic = be), the following example shows that this cannot be the case. In (19) we have a verb followed by a locative as well as a comitative. Both orderings of the postverbal elements are possible without any change in meaning: (19) a.

E ba/*be [ke Zon] [ku-inen mina se] 3sg go home John with=3pl child dem b. E be/*ba [ku-inen mina se] [ke Zon] 3sg go with=3pl child dem home John BOTH: ‘He went to John’s place with those children.’

If the form of the verb depends on the telicity of the predicate, we would expect the same form in (19a) and in (19b). This is not what we find, however. If the verb is followed by a selected element (e.g. locative as in (19a), on a par with (18a)), the form is ba, but when it is followed by a non-selected element (e.g. comitative as in (19a), on a par with (18b)) the form is be. Thus, as in the case of Morisyen, São-Tomense also marks the selectional properties of verbs morphologically. Moreover, in São-Tomense it is also a post-syntactic process. As in Morisyen, if the object is wh-moved to the left periphery of the sentence, the verb shows up be, as shown in (20): (20) Q Andji ku-e subli be/*ba? A E ba/*be losa. Where that=3sg go.up go 3sg go plantation ‘Where did he go up to?’ ‘He went to the plantation.’ Summarizing, the long/short opposition in Morisyen marks the selectional properties of verbs, i.e. whether it is followed by an argument or an adjunct. It parallels São-Tomense where the same grammatical property is marked morphologically (albeit with a suppletion rule). In both cases it concerns a post-syntactic process. The most intriguing about it is that it seems to be independent of the superstrate as well as substrate languages that were present in the (original) language contact situations. Morisyen and São-Tomense neither share a common superstrate nor a

Creole prototypes as basic varieties and inflectional morphology 

set of common substrate languages (cf. Corne 1999; Hagemeijer 2000). As such, we are dealing with a so-called “property of the third kind” (Veenstra 2001), which are arguably the properties particular to creoles alone that can give us insight into the human language faculty.

. Discussion So far, we have shown that the different French-related creole languages have not uniformly (re-) interpreted the “free” formal variation found in verb forms that is due to processes of second language acquisition. Instead, we have identified two different strategies. The long/short opposition in Morisyen (and to some extent in Reunionais and Haitian, see Section 2 above) marks the selectional properties of verbs, i.e. whether it is followed by an argument or an adjunct. In this, the Morisyen pattern differs completely from the Louisianais pattern (and to some extent Reunionais), where there exists a correlation between the long/short opposition and the interpretation of Tense and V-to-I movement. In the remainder of this section we will deal with two questions: (i) in how far can these properties of verbal morphology in French-related creole languages be attributed to the process of second language acquisition and (ii) what conclusions can be drawn for the process of creole genesis. A discussion of the relationship between adult second language acquisition and creole genesis presupposes the specification of commonalities and differences between the two processes. Creole genesis involves – at some point – second language acquisition, and insofar both processes are determined by the human language faculty. There are, however, major differences between the two processes. Véronique (1994: 118) underlines the following: “creolization is a collective endeavour, the identity of the target language is unclear in the case of creolization, and the social matrices involved are dramatically different”. In order to capture the differences one has to take into consideration at least the following factors (cf. Arends 1995; Baker 1982; Baker & Corne 1986; Corne 1994, 1995; Muysken 2001): (21) a. b. c. d. e.

the socio-historic and demographic situation; the input the learner is confronted with; the target towards which the acquisition process is directed; the agents of the process (adults/children); the time dimension of the process.

In adult second language acquisition the agent of the acquisition process is trivially the adult learner endowed with the capacity of processing the L2 input and of gradually building up a series of learner varieties. Untutored adult second language

 Angelika Becker and Tonjes Veenstra

acquisition typically occurs in migration contexts in which the target language is the language of the host community. Most often, the target is a colloquial or dialectal language variety; in any case, it is a fully-fledged language.18 The rate of acquisition and the final stage reached depend upon a number of factors which crucially involve the quantity and quality of the input, the learner’s opportunities of communicating in the second language and personal factors like motivation, the intended length of stay, etc. (cf. Klein 1984). Even in unfavourable acquisition contexts in which there is little contact with target language speakers, the adult learner has to cope with a number of situations in the second language, especially in institutional contexts like social welfare, medical care, municipal authorities, etc. In a favourable acquisition situation (high motivation, possibility of social uprise, contact with native speakers in private life and at the working place, low degree of discrimination, etc.) a learner can approach the competence of a native speaker in the course of a few years (cf. Klein & Dittmar 1979). When the conditions of acquisition are less advantageous, the learner language is likely to fossilize at a lower level of proficiency which is often the Basic Variety. The scenario for creolization is more diverse, and more complex. Presentday French-related creoles have developed in contexts of colonialization and slave trade. They emerged in multilingual communities composed of a dominated group of African slaves (speakers of the substrate languages) and a dominant group of European colonialists (speakers of the superstrate language). In order to assess the role of L2 acquisition in creole genesis the quantitative proportions of these different (African and European) populations have to be taken into account. The assumption that the substrate speakers engaged in a process of second language acquisition is only tenable if one has good reason to believe that they actually could have access to the target language. Baker (1982) was the first to develop a model of the relationship between demographic development and linguistic processes. It is hypothesized that in the beginning of colonization the proportion of target language (superstrate) speakers is sufficient for guaranteeing the substrate population access to the target language up to a point that Baker (1982) calls “Event 1”, i.e. a numerical parity between African and European parts of the population. From that point in time onward the availability of the target language model for L2 learners decreases due to the increasing numbers of slaves being imported into the colony. A new demographic situation is reached at “Event 2” where the number of locallyborn children of the substrate population equals that of the European population. At Event 2 the slave population has nativized to such an extent that creolization may occur. If one accepts the basic ideas behind this model one has to draw at least three conclusions for the process of creole genesis: (i) at an initial stage a process of second language acquisition directed towards the superstrate language has taken place, (ii) at some point in time, access to the superstrate was getting more and more difficult, such that one can assume a “Target Shift” (Baker 1990, 1996),

Creole prototypes as basic varieties and inflectional morphology  G1 G2 G3

L1 L2 L1 L1 L1 (L2

ancestral language(s) Basic Variety (and other approximations of the target language) ancestral language(s) (Post-)Basic Variety (Post-)Basic Variety ancestral language(s))

Figure 1. Multi-generational scenario of Creole genesis

(iii) The process of creolization is driven by two types of agents, namely adults and locally-born children. It, therefore, involves both adult second language acquisition and bilingual first language acquisition. This means that one has to assume a multi-generational scenario. Such a scenario has been first proposed in Corne (1994, 1995) for Tayo and the Indian Ocean creoles (including Morisyen). Further evidence for this approach to creole genesis has been adduced by Roberts (2000) on the basis of careful examination of linguistic evidence, demographic data, and sociological remarks by observers and native-born speakers, for the emergence of Hawai‘i Creole English. The picture is then as in Figure 1. In this scenario, the first generation (G1) consists of immigrants/slaves who pidginize the superstrate (i.e. create a Basic Variety thereof). The next generation (G2) are the locally-born children of G1. According to Corne (1994) this is the crucial generation. As he puts it (1994: 296): Members of G2 retain their ancestral L1 (. . . ); but, like the children of immigrants everywhere, they have to acquire the language of the community as a whole. Their problem is that there is no community L1 apart from pidginized L2 French [Basic Variety in our terms, AB/TV], and they must therefore acquire/create L1 competence on the basis of their exposure to (some subset of) the varieties of pidginized L2 French, variable over time and depending on personal circumstances (. . . ). It is they who participate in the creation of the nascent creole.

Members of the third generation (G3) have lost their ancestral language(s) and are the first monolingual speakers of the newly emerged creole, and are only indirectly exposed to the original ancestral languages due to their being member of a community containing G1 and G2 speakers. We will now interpret some of these basic assumptions in the light of the morphological comparison between French L2 varieties, especially the Basic Variety, and French-related creoles. The question is whether what we know about the acquisition process can help us to further characterize the (different) phases of creole genesis.

 Angelika Becker and Tonjes Veenstra

According to the approach of Klein and Perdue (1997) three levels of second language development can universally be discerned: (i) the Pre-basic Variety, (ii) the Basic Variety and (iii) the Post-basic variety. The structure of the three types of learner varieties is determined by a small number of constraints: pragmatic, semantic and phrasal constraints. The Pre-basic variety is characterized by a lack of verbs, i.e. it is not relevant for the comparison in question. In the Basic Variety the learners’ lexicon is enriched by thematic verbs. The verbs occupy a “middle” position in the utterance: they are preceded by a “subject”-NP and followed by the “object” complements (phrasal constraint).19 Constituent order is semantically and pragmatically constrained: the most agentive argument is assigned the subject role (semantic constraint) and constituents with topic status precede those with focus status (pragmatic constraint). The verbs do not carry functional information. The French learners use both short and long verb forms at this level of proficiency without establishing a functional contrast. Although the structure of the Basic Variety is “simple” in the sense that it is based on a restricted number of (functionally motivated) principles the learner system is communicatively very powerful (cf. Klein & Perdue 1997). This is the reason why the acquisition process often stops at this level. However, the Basic Variety only provides an efficient means of communication as long as its structuring principles coalesce (e.g. agent-topic coalition). The Basic Variety fails in contexts where the principles get into conflict (e.g. when the agent has focus status). In order to deal with such situations of conflict, some learners develop new structural devices, i.e. development beyond the Basic Variety takes place. When the learners go beyond the Basic Variety they gradually approach the target language system. In the domain of verbal morphology they work out a differentiation between finite and infinite verb forms, starting with constructions where the finite verb component, which carries functional information, and the infinite component, carrying lexical information, are structurally separated, i.e. constructions with a finite auxiliary or a finite modal verb and an infinite lexical verb. It is in these contexts that the long forms are first attributed a function, namely that of a past participle or of an infinitive which is in accordance with the target language. Short forms specialize for present tense marking. If we now look at French-related creoles we see two things: (i) the variation between short and long forms is present; (ii) the formal variation corresponds to functional contrasts. These contrasts do not necessarily conform to the oppositions established in French, and furthermore they can vary between different creoles. Thus, the Louisianais and Morisyen patterns mark different (grammatical) properties. The Louisianais (and to some extent Reunionais, see Section 2 above) pattern is quite similar to the pattern that emerges in the Post-Basic Variety. This is not surprising, however, since it is exactly in these colonial contexts that there was an ongoing contact with the superstrate/relexifier language French (this shows that early superstrate-withdrawal is a significant factor in creole genesis). In the case

Creole prototypes as basic varieties and inflectional morphology 

of Louisianais it is a late development, at least two centuries after the start of the settlement (Neumann 1985; Klingler 2000; see Note 13), and not due to Post-Basic Variety developments. With respect to Reunionais, it is only one of the patterns in the language, the other one being more akin to Morisyen (Corne 1999). A comparison between the (Post-) Basic Variety and Morisyen shows two points: (1) there is a correspondance at the level of form, the “long/short” opposition; (2) there is no correpondance at the level of form-function assignment: in the Basic Variety, the formal variation is non-functional, and the form-function attribution which occurs in the Post-basic Variety differs considerably from that found in Morisyen (and presumably Haitian and Reunionais to some extent, see Section 2 above). The functional opposition worked out by second language learners are targeted towards the variety of French the learners encounter as input. The functional differentiation in Morisyen is of a completely different nature than the one given in French. This leads to the following picture. In creole genesis, we have an initial process of adult second language acquisition (by G1) directed towards the target language (TL) French. The level of attainment was certainly diversified between learners,20 but it can plausibly be assumed that the Basic Variety was well represented. When the early settlement system changed to a plantation system and the number of imported slaves increased, the possibility of access to the target language French receded. French was no longer available as a target language and therefore a targetshift occurred. We take the existence of the long-short variation in creoles as an indication that the new target was the Basic Variety. The formal variation in creole languages can then be seen a “Basic Variety continuity”. French can not longer have been the target language as the functions ascribed to the long/short variation in some creoles are independent of French.The target shift, therefore, must have taken place before the formal variation acquired a function. In other words, the opposition in Morisyen acquired its function after the target shift. It is our contention that it was G2 that fitted the formal variation into a paradigm, marking a basic distinction at the level of Conceptual Structure (argument vs. adjunct) on the basis of their innate language faculty. What is common, then, to the process of second language acquisition and the process of creolization is a development that leads to the Basic Variety. The crucial difference between the two processes is the occurrence of a Target Shift which leads to the Basic Variety being accepted as a separate norm.21 From then on, the two processes take different courses. In second language acquisition the expansion of the Basic Variety leads to Post-basic varieties approaching the target language. In creolization, on the other hand, the expansion of the Basic Variety takes place in two steps. The different routes are schematically represented in Figure 2. Members of G1 continue to expand the Basic Variety by different processes of grammaticalization, in particular paradigmatization and syntacticization, resulting

 Angelika Becker and Tonjes Veenstra Basic Variety formation similar in both cases Basic Variety syntactic, semantic & pragmatic constraints

SLA: ongoing access to the TL, therefore no TL-shift

Creole: reduced access to the TL, therefore TL-shift: Superstrate > Basic Variety Basic Variety expansion

SLA approaching the TL

Creole expansion of the TL (part I) [by G1: adults] expansion of the TL (part II) [by G2: children]

Figure 2. Differences and similarities in the expansion phase

from conflicting Basic Variety-principles, constrained by cognition and, possibly, L1-knowledge. Members of G2, who are bilingual in both the Basic Variety and the ancestral (substrate) language(s), also expand the newly emerging community language. In this generation cross-linguistic interference plays a role, i.e. structures get more pronounced when there is structural similarity between the language pairs in bilingual situations (convergence, cf. Kouwenberg 2001). It is also this generation that in our view is responsible for some of the relexificational effects that have been documented by Lefebvre (1998) and Lumsden (1994, 1995, 1999a, b) for Haitian. With this generation the language shift starts and this shift is ended by G3, for whom the newly emerged creole language is their only mother tongue.

Notes . This does not mean that these languages are in any way to be regarded “simple”, however. First, such a statement only makes sense if one has a theory of complexity to fall back on (cf. Muysken 1988; Veenstra 1996). Second, the morphological complexity of a language need not be defined over the inflectional domain solely, because it can also reside in the derivational component (cf. the extensive discussion in DeGraff 2001), or be due to different compounding processes (e.g. many creoles have used such processes to reconstitute functional

Creole prototypes as basic varieties and inflectional morphology 

elements such as question words (Muysken & Smith 1990) and reflexives (Muysken 1993; Muysken & Smith 1995; Veenstra 1996). . The project – Second Language Acquisition of Adult Immigrants – was carried out in five European countries (France, Germany, Great Britain, the Netherlands, and Sweden) between 1981 and 1988. It was supported by the European Science Foundation and coordinated by the Max-Planck-Institute for Psycholinguistics in Nijmegen, the Netherlands. A description of the project design is given in Perdue (1993). . Five target languages were combined with two source languages each: (1) Dutch – Turkish/Moroccan Arabic, (2) English – Italian/Punjabi, (3) French – Spanish/Moroccan Arabic, (4) German – Italian/Turkish, (5) Swedish – Spanish/Finnish. . In this we depart from the classical Bickertonian view of creolization as a catastrophic one-generation process (1981, 1984) as well as the gradualist model of creolization, in which the jelling of the newly emerging language takes place over many (i.e. more than three) generations (Arends 1993; Arends & Bruyn 1995; Carden & Stewart 1988; Singler 1986, 1990). . In this we depart from the relexification hypothesis of Lefebvre and Lumsden (1994) in which the adults are the agents of the relexification process. Our proposal is more in line with Muysken’s (1981, 1986) original observations regarding the emergence of Media Lengua in a bilingual context. Furthermore, contrary to Lefebvre (1998: 34) and Plag (2001: 177), relexification is not a well-known nor a well-documented phenomenon from studies of (adult) untutored second language acquisition in immigration contexts. We do find comparable processes in bilingual acquisition, however, see e.g. Schmitt (2000) on “covert” codeswitching in immigrant children from Russia, where she reports on different kinds of (in our terms) relexificational effects. . The imparfait and the passé composé express past time reference with the aspectual opposition “imperfective” vs. “perfect”. . The long forms can appear with other affixes depending on the verb forms in the input the learner is confronted with. The dominance of the V-/e/ pattern is probably due to the frequency of -er-verbs in the target language. The long forms of the target language -ir-verbs mostly correspond to the standard French past participle form. For example, the learners use long forms like /sorti/ (Fr. sorti ‘gone out’) or /parti/ (Fr. parti ‘left’). . The hyphen in the examples of learner varieties indicates that the expression is used as an unanalyzed root form. . This does not mean, however, that temporal relations are not or cannot be specified in the Basic Variety. Temporal relations are established by making use of lexical and pragmatic devices. The learners have quite a rich repertoire of temporal adverbials (e.g. calendaric adverbials, anaphoric adverbials like après, deictic adverbials like maintenant) which allows them to indicate the time span relative to which they want to make an assertion. In narratives they can also profit from a principle of text organization, the “principle of natural order” according to which the order of mention of events corresponds to the order in which they occurred. . Moroccan Arabic marks aspect and aktionsart categories on the verb by means of verbal prefixes, suffixes and infixes (cf. Noyau et al. 1995: 185).

 Angelika Becker and Tonjes Veenstra . The term “proto-auxiliary” relates to the fact that these forms do not take all the features of fully developed auxiliaries, i.e. person and agreement features (cf. Starren 2001: 160). . Class II verbs are by and large morphologically invariant (i.e. only exhibiting a long form). Rottet (1992: 268) argues that in some cases verbs of this class could best be considered as class I verbs in disguise, having both an inflected and an uninflected form. This opposition is not morphologically marked as with “real”class I verbs, but rather shows up in the correlation between the temporal interpretation of the sentence and the position in which the verb appears, a pattern which we find with “real” class I verbs (see below in the main text). Most other verbs in this class do not exhibit this pattern, however (cf. Neuman 1985; DeGraff 1997). . To be more precise, Klingler (2000) distinguishes two main varieties of Louisianais: Mississippi Creole and Teche Creole. He argues that Teche Creole once shared with nineteenth-century Mississippi Creole a system of invariant, one-stem verbs and that the system of two-stem verbs found in Teche Creole today is a recent innovation most likely attributable to decreolization through contact with Cajun French. (2000: 29) Thus, the long/short opposition is only found in those varieties that are in prolonged contact with French. . In the first attestations of Morisyen (1749), there is no distinction between short and long forms of verbs (only long forms attested). Short forms are first found, unsystematically, in Pitot (1805). In the late-nineteenth century (the texts in Baissac 1880) both the set of variable verbs and the frequency with which they adopt their short forms is broadly the same as in modern Morisyen (cf. Corne 1982). . In Haitian, the long/short contrast in verb forms also appears to play a role to distinguish between the transitive and intransitive/middle/anti-causative/passive forms of certain predicates, as in (i): (i)

a. Mwèn fè/*fèt kabann lan. b. Kabann lan fèt/*fè. 1sg make bed det bed det make ‘I’ve made the bed.’ ‘The bed is made.’

Massam (1989) and DeGraff (2001) indicate there are certain predicates that undergo this alternation (in addition to fè/t ‘make’, are enkye/te ‘worry’ and kowonp/i ‘corrupt/corrupted’ the only attestations we found so far in the literature). Most verbs in Massam (1989) that occur in anti-causative, middle and/or passive constructions do not show the long/short opposition, however. . An implicit argument is the suppressed external argument of the verb which plays a role in the interpretation of verbal passives without by-phrases. There are a number of reasons to assume that such an implicit argument is represented syntactically (cf. Roeper 1987 for English; Verrips 1996 for Dutch; Veenstra 1998 for Saramaccan). The empirical core of the argument is that the implicit argument can control the non-overt PRO subject of an infinitival purpose clause. This is illustrated in (i): (i)

The glass was broken [pro to hurt Mary].

Creole prototypes as basic varieties and inflectional morphology 

Furthermore, adverbial phrases can predicate over the implicit argument: (ii) a. b.

The glass was broken on purpose The glass was broken with a hammer.

The Agent-oriented adverbial in (ii-a) modifies the implied ‘breaker’, not the glass. The instrumental prepositional phrase in (ii-b) is interpreted as the instrument used by the ‘breaker’. Arguably, there is a PRO in these adverbial phrases too, controlled by the implicit argument. . The relevant data are given below: (i)

a.

b.

(ii) a.

b.

(iii) a.

b.

Ban brans i n kase ek divan. plur branch 3sg asp break with wind ‘The branches are/have been broken by the wind.’ Ban brans i n kas ek divan. plur branch 3sg asp break with wind ‘The branches broke with/because of the wind.’ Nuvel pe fane dan lari. news asp spread loc street ‘The news is being spread in the streets.’ Nuvel pe fan dan lari. news asp spread loc street ‘The news is spreading in the streets.’ duri vande dan labutik. rice sell loc shop ‘Rice is sold in shops.’ Duri van dan labutik. rice sell loc shop ‘Rice sells in shops.’

passive

anti-causative

passive

anti-causative

passive

middle

We leave a full treatment of this issue for future research. . This is, of course, a simplified picture as the input can also contain foreigner talk or learner varieties of the target language. In the vast majority of cases, however, a learner will encounter some standard/ colloquial/ dialectal target language input e.g. by mass media. . Criteria for subject-hood and object-hood are derived solely from the referent’s semantic relation with the predicate and the pragmatic status of the constituent. The subject is determined on the basis of agentivity and background status, the object on the basis of nonagentivity and focus status. Morphological case marking occurs only at very late stages of acquisition. . There also must have been learners around with closer approximations of the competence of “native” speakers in order to account for the continuities of French morphology in Haitian as documented and discussed extensively by DeGraff (2001). This implies that the creole continuum (Bickerton 1975; Rickford 1987) must have developed in the colony from quite early on. Early superstrate-withdrawal also plays an important role to account for the

 Angelika Becker and Tonjes Veenstra

difference between the situation in Surinam (early withdrawal and no continuum) and the Jamaican and Guyanese situation (creole continuum and no early withdrawal). . As Muysken (2001: 160) notes: before a pidgin can be creolized, it must need to be accepted as a separate norm. The crystallization of the pidgin and the stability of a pidginized variety may be furthered by: (i) a socially homogeneous community speaking the pidginized variety; (ii) reciprocal native language / target language community communication in the pidginized variety; (iii) use of the pidginized variety in communication between different groups of non-target language speakers; (iv) the variety is learned by newcomers as a system; (v) pidginized variety mutually intelligible with target language.

References Arends, J. (1993). Towards a gradualist model of creolization. In F. Byrne & J. Holm (Eds.), Atlantic meets Pacific (pp. 371–380). Amsterdam & Philadelphia: John Benjamins. Arends, J. (1995). The socio-historical background of creoles. In J. Arends, P. Muysken, & N. Smith (Eds.), 15–24. Arends, J., Muysken, P., & Smith, N. (Eds.). (1995). Pidgins and Creoles: An introduction. Amsterdam & Philadelphia: John Benjamins. Arends, J. & Bruyn, A. (1995). Gradualist and developmental hypotheses. In J. Arends, P. Muysken, & N. Smith (Eds.), 303–314. Baissac, C. 1880). Etude sur le patois créole mauricien. Nancy: Imprimerie Berger-Levrault. Baker, M. (1988). Incorporation. Chicago: Chicago University Press. Baker, M., Johnson, K., & Roberts, I. (1989). Passive arguments raised. Linguistic Inquiry, 20, 219–252. Baker, P. (1982). On the origins of the first Mauritians and of the creole language of their descendants. A refutation of Chaudenson’s ‘Bourbonnais’ theory. In P. Baker & C. Corne (Eds.), Isle de France Creole: affinities and origins (pp. 131–259). Ann Arbor: Karoma. Baker, P. (1990). Off Target? Journal of Pidgin and Creole Languages, 5(1), 107–119. Baker, P. (1996). Pidginization, creolization, and français approximatif. [Review article of Chaudenson] Journal of Pidgin and Creole Languages, 11(1), 95–120. Baker, P. & Corne, C. (1986). Universals, Substrata and the Indian Ocean Creoles. In P. Muysken & N. Smith (Eds.), Substrata versus Universals in Creole Genesis (pp. 163–183). Amsterdam & Philadelphia: John Benjamins. Baker, P. & Syea, A. (1991). On the copula in Mauritian Creole, past and present. In F. Byrne & T. Huebner (Eds.), Development and Structures of Creole Languages (pp. 159–175). Amsterdam & Philadelphia: John Benjamins. Bickerton, D. (1975). Dynamics of a creole system. Cambridge: Cambridge University Press. Bickerton, D. (1977). Change and variation in Hawaiian English. Vol. 2: Creole syntax. University of Hawaii: Social Science and Linguistics Institute.

Creole prototypes as basic varieties and inflectional morphology 

Bickerton, D. (1981). Roots of Language. Ann Arbor: Karoma. Bickerton, D. (1984). The language bioprogram hypothesis. The Behavioral and Brain Sciences, 7, 173–188. Borer, H. (1984). The projection principle and rules of morphology. NELS, 14, 16–33. Umass publication. Carden, G. & Stewart, W. (1988). Binding Theory, Bioprogram, and Creolization: Evidence from Haitian Creole. Journal of Pidgin and Creole Languages, 3(1), 1–68. Corne, C. (1981). A re-evaluation of the predicate in Ile-de-France Creole. In P. Muysken (Ed.), Generative Studies on Creole Languages (pp. 103–129). Dordrecht: Foris. Corne, C. (1982). A contrastive analysis of Reunion and isle de France Creole French. In P. Baker & C. Corne (Eds.), Isle de France Creole: affinities and origins (pp. 7–129). Ann Arbor: Karoma. Corne, C. (1994). Relativization and thematization in Tayo and the implications for creole genesis. Journal of Pidgin and Creole Languages, 9(2), 283–304. Corne, C. (1995). Nana k nana, nana k napa: The paratactic and hypotactic relative clauses of Reunion Creole. Journal of Pidgin and Creole Languages, 10(1), 57–76. Corne, C. (1999). From French to Creole. London: University of Westminster Press. Cornelis, L. & Verhagen, A. (1995). Does Dutch really have a passive? In M. Den Dikken & K. Hengeveld (Eds.), LIN 1995 (pp. 49–60). Amsterdam & Philadelphia: John Benjamins. DeGraff, M. (1997). Verb syntax in creolization (and beyond). In L. Haegeman (Ed.), The new comparative syntax (pp. 64–94). London: Longman. DeGraff, M. (Ed.). (1999). Language Change and Creation: creolization, diachrony and development. Cambridge, MA: MIT Press. DeGraff, M. (2001). Morphology in creole genesis: a prolegomenon. In M. Kenstowicz (Ed.), Ken Hale: a life in language (pp. 53–122). Cambridge, MA: MIT Press. Dietrich, R., Klein, W., & Noyau, C. (Eds.). (1995). The acquisition of temporality in a second language. Amsterdam & Philadelphia: John Benjamins. Fromkin, V. (Ed.). (2000). Linguistics. London: Blackwell. Hagemeijer, T. (2000). Serial Verb Constructions in São-Tomense. MA Thesis, Universidade de Lisboa, Portugal. Harris, M. (1987). French. In B. Comrie (Ed.), The World’s Major Languages (pp. 210–235). London/ New York: Routledge. Klein, W. (1984). Zweitspracherwerb. Königstein/Ts.: Athenäum. Klein, W. & Dittmar, N. (1979). Developing Grammars. Berlin etc.: Springer. Klein, W. & Perdue, C. (Eds.). (1992). Utterance Structure. Developing grammars again. Amsterdam/ Philadelphia: John Benjamins. Klein, W. & Perdue, C. (1997). The basic variety. Second Language Research, 13(4), (pp. 301– 347). Klingler, T. (2000). Louisiana Creole: the multiple geneses hypothesis reconsidered. Journal of Pidgin and Creole Languages, 15(1), 1–36. Kouwenberg, S. (2001). Convergence and explanations in creole genesis. In N. Smith & T. Veenstra (Eds.), Creolization and Contact (pp. 219–243). Amsterdam & Philadelphia: John Benjamins. Lefebvre, C. (1998). Creole genesis and the acquisition of grammar. Cambridge: Cambridge University Press.

 Angelika Becker and Tonjes Veenstra

Lefebvre, C. & Lumsden, J. (1994). Relexification in Creole Genesis. Université du Québec à Montréal. Paper read at the MIT Symposium on the Role of Relexification in Creole Genesis: The Case of Haitian Creole. Research report edited by C. Lefebvre & J. Lumsden. Montreal. Levin, B. & Rappoport, T. (1986). The formation of adjectival passives. Linguistic Inquiry, 17, 623–661. Lumsden, J. (1994). Possession: substratum semantics in Haitian Creole. Journal of Pidgin and Creole Languages, 9, 25–50. Lumsden, J. (1995). Aspect and lexical semantics in Haitian Creole. The Linguistic Review, 12, 123–142. Lumsden, J. (1999a). Language Acquisition and Creolization. In M. DeGraff (Ed.), Language Change and Creation: creolization, diachrony and development (pp. 129–158). Cambridge, MA: MIT Press. Lumsden, J. (1999b). The Role of Relexification in Creole Genesis. Journal of Pidgin and Creole Languages, 14(2), 225–258. Macías, R. (2000). The flowering of America: linguistic diversity in the United States. In S. McKay & S. C. Wong (Eds.), New Immigrants in the United States (pp. 11–57). Cambridge: Cambridge University Press. Massam, D. (1989). Predicate argument structure in Haitian Creole. Revue de Quebecoise Linguistique, 95–130. McWhorter, J. (1998). Identifying the creole prototype: vindicating a typological class. Language, 74, 788–818. Mufwene, S. (1990). Transfer and the substrate hypothesis in creolistics. Studies in Second Language Acquisition, 12, 1–23. Muysken, P. (1981). Halfway between Quechua and Spanish: the case for relexification. In A. Highfield & A. Valdman (Eds.), Historicity and Variation in Creole Studies (pp. 52–87). Ann Arbor: Karoma. Muysken, P. (1986). The Nature of Pidgins and Creoles in Space, Time, and Society. Paper presented at CUNY, New York, July. Muysken, P. (1988). Are Creoles a Special Type of Language. In F. Newmeyer (Ed.), Linguistics: the Cambridge Survey, II (pp. 285–301). Cambridge: Cambridge University Press. Muysken, P. (1993). Reflexes of Romance verb + clitic combinations in Papiamentu. In F. Byrne & D. Winford (Eds.), Focus and grammatical Relations in the Creole Languages (pp. 285–301). Amsterdam & Philadelphia: John Benjamins. Muysken, P. (1996). Verdwijnende talen staan samen sterker. de Volkskrant, 23, January 13, 1996. Muysken, P. (2000). Semantic transparency in Lowland Ecuadorian Quechua MorphoSyntax. Linguistics, 38, 973–988. Muysken, P. (2001). The origin of creole languages: The perspective of second language learning. In N. Smith & T. Veenstra (Eds.), Creolization and Contact (pp. 157–174). Amsterdam & Philadelphia: John Benjamins. Muysken, P. & Law, P. (2001). Creole Studies. A theoretical linguist’s field guide. GLOT International, 5(2), 47–57.

Creole prototypes as basic varieties and inflectional morphology 

Muysken, P. & Smith, N. (1990). Question words in pidgin and creole languages. Linguistics, 28(4), 883–903. Muysken, P. & Smith, N. (1995). Reflexives. In J. Arends, P. Muysken, & N. Smith (Eds.), 271–288. Neumann, I. (1985). Le Créole de Breaux Bridge, Louisiane: étude morphosyntaxe, textes, vocabulaire. Hamburg: Helmut Buske. Noyau, C., Et-Tayeb Houdaïfa, Vasseur, M.-T., & Véronique, D. (1995). The acquisition of French. In R. Dietrich, W. Klein, & C. Noyau (Eds.), The acquisition of temporality in a second language, 145–209. Perdue, C. (Ed.). (1993). Adult language acquisition: Crosslinguistic perspectives. Vol. 1 Field methods, Vol. 2 The results. Cambridge: Cambridge University Press. Plag, I. (2001). Review of Lefebvre (1998). Journal of Linguistics, 36(1), 176–180. Pollock, J.-Y. (1989). Verb movement, Universal Grammar, and the structure of IP. Linguistic Inquiry, 20, 365–424. Radford, A., Atkinson, M., Britain D., Clahsen, H., & Spencer, A. (1999). Linguistics: an introduction. Cambridge: Cambridge University Press. Rickford, J. (1987). Dimensions of a creole continuum. History, texts, & linguistic analysis of Guyanese Creole. Stanford: Stanford University Press. Roberts, S. (2000). Nativization and the genesis of Hawaiian Creole. In J. McWhorter (Ed.), Language Change and Language Contact in Pidgins and Creoles (pp. 257–300). Amsterdam & Philadelphia: John Benjamins. Roeper, T. (1987). Implicit arguments and the head-complement relation. Linguistic Inquiry, 18, 267–310. Rooy, V. de (1995). Shaba Swahili. In J. Arends, P. Muysken, & N. Smith (Eds.), 179–190. Rooy, V. de (1996). Cohesion through Contrast. Ph.D. University of Amsterdam. Rottet, K. (1992). Functional categories and verb raising in Louisiana Creole. Probus, 4, 261– 289. Scalise, S. (1984). Generative Morphology. Dordrecht: Foris. Schmitt, E. (2000). Overt and covert codeswitching in immigrant children from Russia. The International Journal of Bilingualism, 4(1), 9–28. Selinker, L. (1972). Interlanguage. International Review of Applied Linguistics, 10, 209–231. Seuren, P. (1990). Verb syncopation and predicate raising in Mauritian Creole. Linguistics, 28(4), 809–844. Siegel, J. (1999). Transfer constraints and substrate influence in Melanesian Pidgin. Journal of Pidgin and Creole Languages, 14(1), 1–44. Singler, J. (1986). Short note. Journal of Pidgin and Creole Languages, 1, 141–145. Singler, J. (1990). On the use of sociohistorical criteria in the comparison of creoles. Linguistics, 28(4), 645–659. Spencer, A. (1991). Morphological Theory. London: Blackwell. Starren, M. (2001). The Second Time. Ph.D. Catholic University of Brabant. Syea, A. (1992). The short and long form of verbs in Mauritian Creole: Functionalism versus formalism. Theoretical Linguistics, 18(1), 61–97. Thomason, S. (2001). Contact-induced language change and pidgin/creole genesis. In N. Smith & T. Veenstra (Eds.), Creolization and Contact (pp. 249–262). Amsterdam & Philadelphia: John Benjamins.

 Angelika Becker and Tonjes Veenstra

Veenstra, T. (1996). Serial Verbs in Saramaccan: predication and creole genesis. The Hague: HAG. Veenstra, T. (1998). The syntax of passives in Saramaccan: architectural implications. Unpublished manuscript, University of Potsdam. Veenstra, T. (2001). Properties of the Third Kind in Language Contact. Paper presented at the ACLC Vrijdagclub, University of Amsterdam, May 18. Veltman, C. (2000). The American linguistic mosaic: understanding language shift in the United States. In S. McKay & S. C. Wong (Eds.), New Immigrants in the United States (pp. 58–94). Cambridge: Cambridge University Press. Véronique, D. (1994). Naturalistic adult acquisition of French as L2 and French-based creole genesis compared: insights into creolization and language change? In D. Adone & I. Plag (Eds.), Creolization and Language Change (pp. 117–173). Tübingen: Niemeyer. Verrips, M. (1996). Potatoes must Peel. The Hague: HAG. Wasow, T. (1977). Transformations and the lexicon. In P. Cullicover, T. Wasow, & A. Akmajian (Eds.), Formal Syntax (pp. 327–360). New York: Academic Press.

JB[v.20020404] Prn:29/04/2003; 15:31

P II

Anaphoric relations

F: SIB26P2.tex / p.1 (36-61)

JB[v.20020404] Prn:26/06/2003; 13:29

F: SIB2610.tex / p.1 (35-99)

Information structure in narratives and the role of grammaticised knowledge A study of adult French and German learners of English Mary Carroll and Monique Lambert

.

Introduction

The present study of advanced adult learners of a second language deals with factors which determine information structure in narratives, focusing on the extent to which adult learners succeed in acquiring the principles of information structure of the target language. Information structure is treated here in a broad sense and encompasses factors which determine the kind of information which is selected for expression and how it is mapped into form. The study is part of a series of crosslinguistic comparisons which examine the role of grammaticised meanings in facilitating the types of decisions required of speakers when organising information for expression in tasks such as narratives or object descriptions (von Stutterheim & Lambert, in press; Carroll, Murcia-Serra, Watorek, & Bendiscioli 2000). Languages exhibit considerable diversity in their coding preferences in that meanings which are coded in some languages in lexical form are mapped into grammatical form in others. Grammaticised meanings play a crucial role in determining the decisions which speakers make in information structure, as indicated by previous cross-linguistic studies. So grammatical form is not viewed in the present context as an autonomous system independent of meaning, but as one which embodies a system of meanings that is treated in a given language as particularly important or essential in the interpretation and conceptualisation of reality (cf. Wierzbicka 1995). As stated in Hockett (1954), cross-linguistic diversity consists less in what is possible to specify than what is relatively easy or hard to specify (cf. Talmy 1987; Slobin 1991, 1996; Berman & Slobin 1994). There are meanings which may be “easier” to specify in context in the sense that they are coded more frequently by speakers of a given language, as, for example, the expression of a

JB[v.20020404] Prn:26/06/2003; 13:29

F: SIB2610.tex / p.2 (99-171)

 Mary Carroll and Monique Lambert

temporal concept such as ongoingness (they are running). Speakers of languages in which a concept of this kind is grammaticised on the verb (English, Spanish, Arabic, for example) are more likely to express corresponding aspects of a dynamic situation, when relevant in context, compared to speakers who have to resort to lexical means in order to convey the same kind of information, such as in French or German (cf. von Stutterheim & Nuese, forthcoming). The present analysis focuses on the nature of language-specific systems of grammaticised meanings, and how such systems drive decisions at the level of macro- and micro-planning in narratives. At the level of macro-planning speakers have to deal with a range of tasks: information selection, that is, “considering what to say” (Levelt 1996), and its segmentation into propositional units. Further decisions which involve both macro- and micro-planning concerns the mapping of propositional information at clause level (main or subordinate clause), specification of relations in linking or “linearising” propositional units (temporal, causal relations), selection of means in referring to entities, for example. In narratives, where events play an essential role, these decisions cover information which spans the different conceptual domains that constitute core elements of a dynamic situation, world, time, space, entities such as persons and things, events (cf. von Stutterheim 1997). In current models of language production (Levelt 1996, 1999) it is assumed that language-specific knowledge comes into play at the level of micro-planning only, and does not determine decisions relating to steps in macroplanning. In contrast to these models, the cross-linguistic comparisons of information organisation and information structure in descriptions and narratives provide clear evidence of language-specific principles at the level of macro-planning (cf. Section 2.1). The implications for second language acquisition are relevant since language-specific principles which relate to macro-planning seem very difficult to acquire. The principles at issue will be illustrated by a summary of a study on descriptions which demonstrate the role of a system of grammaticised meanings and how they affect decisions on global terms with respect to information structure, showing (a) how grammaticised forms work in unison to promote coherence in information structure, and (b) how second language learners deal with these factors. The findings prompted the present cross-linguistic study on narratives which examines how advanced French and German learners of English structure information, comparing the principles found with those in narratives of native speakers of English, French, German, as well as Italian and Spanish. The comparison between descriptions and narratives illustrates, for example, how a grammatical category such as the syntactic subject, and the different ways in which it is grammaticised across the different languages, plays a central role both in information selection and information structure in both tasks. So we will start by summarising the findings for descriptions.

JB[v.20020404] Prn:26/06/2003; 13:29

F: SIB2610.tex / p.3 (171-234)

Information structure in narratives and the role of grammaticised knowledge 

. Information structure in descriptions The cross-linguistic study of object descriptions analyses the principles governing information structure in texts of advanced learners of German (L2 English– German, L2 Spanish–German), comparing them with those found in L1 German, English, French, Italian, and Spanish. The results show that there are clusters of grammaticised meanings which determine both information selection and information structure. Principles found for English compare to those observed in the Romance languages, and are clearly associated with grammatical features relating to the syntactic subject, word order constraints, and the morphosyntactic structure of spatial expressions. In the Romance languages and English, categories of information which have a high potential in assuming the role of topic are those which map into the syntactic subject (Givón 1983; Lambrecht 1994). This grammatical category channels the decisions which speakers make in information selection and topic focus assignment. In the case of object descriptions (picture depicting an old town centre), the type of information which maps into the syntactic subject is, at the highest level, the picture itself. In the descriptions of speakers of English and the Romance languages, attention is directed to individual entities as a feature of the picture by means of existentials (there is a square with a fountain). The syntactic subject occurs in expletive form (there is ...), and utterances coded in this way constitute statements about the picture: there is ... (topic: picture) a square (focus or new information, with topic used in the sense of background).1 Unlike in English, topic status is not assigned to the picture in German, but to the set of spaces or places which constitute the area in question (vorne im Bild ist eine Strasse,’at the front of the picture is a street’; auf der linken Seite ist eine Apotheke, ‘on the left hand side is a drugstore’). The picture is divided into spatial regions, and entities are grouped and selected for mention on the basis of their location. 18 out of 20 speakers of German select this option, while only 2 out of 20 select the other option and assign topic status to the picture, using existentials (es gibt ein x, ‘there is an x’) as in English and the Romance languages, and not locationals (auf dem Platz ist ein Brunnen, ‘on the square is a fountain’). The different principles established are also systematically reflected in the types of spatial concepts used to divide the picture globally into sections and the types of expressions used in reference maintenance. The preferred option chosen in German can be attributed to the following grammatical features, which are also relevant for information structure in narratives: on the one hand, there is the ‘verb second’ constraint in main clauses, whereby finite verbs form the second main constituent. The V2 constraint creates a ‘slot’ or “Vorfeld” (forefield) in main clauses, which, significantly, can accomodate only one constituent. Categories of information and syntactic constituents which map into the “Vorfeld” are prime candidates in the assignment of topic status in in-

JB[v.20020404] Prn:26/06/2003; 13:29

F: SIB2610.tex / p.4 (234-253)

 Mary Carroll and Monique Lambert

formation structure. Since word order is relatively free, however, the syntactic subject is not confined to preverbal position and can therefore also be used in coding focus or new information, as indicated in the examples above (at the front of the picture is a street = focus or new information). If the syntactic subject is assigned this latter function, there has to be information and constituents (adverbials or prepositional phrases) with topic status which can then systematically fill the preverbal ‘slot’ or “Vorfeld”. As will be shown below, this factor is also crucial in information structure in narratives, and the type of information which can systematically fill this slot in a narrative context is not spatial but temporal (und dann passiert x, ‘and then happens x’). The V2 constraint constitutes one of the significant structural differences between German and the Romance languages, along with English, in information structure. As mentioned above, speakers of German prefer to assign topic status to spaces and places on the picture in object descriptions and so can use the syntactic subject to code new or focus information, an option which is also supported by the morphosyntactic structure of spatial expressions in German. Speakers of German and the Romance languages thus show different sets of preferences in information structure and the preferred options consist of uniform clusters of form function relations, which are clearly supported by grammaticised features of the languages. The language-specific ‘systems of meanings’ allow speakers to code information spanning different domains in consistent terms (cf. Carroll, Murcia-Serra, Watorek, & Bendiscioli 2000). The clusters observed in information structure reflect fundamental differences in the conceptual organisation and representation of information for expression. When the spaces which make up the object are assigned topic status, for example, entities are selected for mention on the basis of the region of space at which they are located, (e.g. mention the things in the centre of the picture). When the picture itself is assigned topic status via the syntactic subject, attention is directed to entities as features of the picture by means of existential statements. They are often grouped for mention in terms of factors such as function, size, etc., and not necessarily the location which they share. The comparison of these findings with principles found in the learner varieties of very advanced learners of German (L2 English–German, L2 Spanish–German) show that the learners retain core principles of their source languages in information structure. The learning problem at advanced stages of learning is not one of linguistic form, the individual grammatical means of the target language German have been acquired. The problem for the learners is they have not yet fully recognised the language-specific system of meanings or clusters of form function relations which determine the preferred information structure in the target language. The coding options selected indicate that learner varieties, along with the source and target languages, also strive toward a unifying set of principles which ensure consistency in information selection and information structure across a complex range of decisions, thereby supporting text coherence at all levels. The

JB[v.20020404] Prn:26/06/2003; 13:29

F: SIB2610.tex / p.5 (253-307)

Information structure in narratives and the role of grammaticised knowledge 

findings indicate that the principle of consistency in information structure may impede acquisition of the target language norm (cf. Carroll, Murcia-Serra, Watorek, & Bendiscioli 2000).

. Information structure in narratives The present study of information selection and information structure shifts its focus to narratives and examines the principles found in narratives of French and German learners of English. This analysis of the learner varieties is based on a wider comparison of information structure in narratives of native speakers of English, French, German, Italian, and Spanish (cf. Carroll, Lambert, & Murcia-Serra, forthcoming). The same stimulus, a film re-telling, was used with all language groups. The categories studied cover both information selection, that is, the type of information which speakers find worthy of mention in re-telling the content of the film, and the coding options which speakers select when mapping information into form. The data base includes 20 native speakers in each language group and consists mainly of university students. They were all asked to re-tell the content of a silent film (‘what happened’?) under similar conditions. The film relates the adventures of a clay figure who wakes up in a desert-like world and sets out on a search for water. His quest takes him through four different worlds (a desert world, a paper world, a stone world, and finally a world dominated by robots and other industrial machines). The French and German learners of English are also students or post graduates, and have all spent time in an English-speaking country, ranging from six months to two years. The German learners of English are for the most part advanced, while the French learners of English include both advanced as well as some near-native speakers. All narratives were told in the present tense, even though they were asked to tell ‘what happened’? The narratives of the native speakers average 180 utterances in length while those of the learners is 150.

. Information selection – deciding “what to say” There are basically two types of information that can be distinguished in the film: (a) information which involves the protagonist and what he does and (b) information concerning inanimate or environmental elements and forces such as wind, sand, water, rocks, and finally at the end of his quest, different kinds of industrial machines.

JB[v.20020404] Prn:26/06/2003; 13:29

F: SIB2610.tex / p.6 (307-363)

 Mary Carroll and Monique Lambert

The analysis starts with a comparison of the extent to which speakers of English, German, and the Romance languages relate to the two types of entities (person, things). The decision to include information selection in the analysis was prompted by the studies of spatial descriptions, as discussed above, and the role of the syntactic subject in English and the Romance languages in directing attention to entities in information structure. The findings for information selection in the narratives show similar coalitions in that there are significant differences between German, on the one hand, and English and the Romance languages, in the extent to which attention is directed to both types of entities – the protagonist and inanimate elements. Speakers of languages in which the subject plays a principal role in information structure, in this case English, French, Italian, and Spanish, refer not only to the activities of the protagonist in the retelling, but also relate to environmental elements in their own right, showing a significantly higher number of mentions compared to speakers of German. References to inanimate entities were counted irrespective of the way in which they are mapped into form (as syntactic subject main clause, syntactic subject of subordinate clause, underyling subject in a passive construction). Example 1. References to inanimate entities in the narratives inanimate entity: the pieces of paper keep flying past him protagonist: so he quickly gets out of the way References to inanimate entities amount to 34.5 in English, 35.8 in French, 37.1 in Italian, 30.9 in Spanish but there is a significant drop in German 24.5 (average % values for 20 speakers). The difference between German and the other languages is statistically significant: t-test English–German p = 0.006 highly significant; Spanish–German p = 0.04 significant; while English–Spanish p = 0.25 is not significant; English–French p = 0.21 is also not significant. German speakers focus attention in their narratives on the protagonist, often omitting reference to prominent inanimate entities. One scene in which the protagonist (a clay figure) is exposed to an inanimate element, for example, is the encounter with a huge rock which is dropping from the sky. 85.7% of the speakers of English refer to this incident, while only 40% of the German speakers do so, differences which are also statistically significant. The factor which determines whether speakers of German relate to such forces or not is related to the protagonist: in the German narratives forces are consistently mentioned only when they have clear consequences for the protagonist. The rock in the scene mentioned misses the protagonist and is thus not worthy of mention to the degree observed in the other languages. In German the attention directed to environmental elements is filtered, so to speak, by selection principles which are associated with the protagonist. The cross-linguistic differences hold across a series of scenes and are described in detail in Carroll, Lambert, Murcia-Serra, forthcoming.

JB[v.20020404] Prn:26/06/2003; 13:29

F: SIB2610.tex / p.7 (363-439)

Information structure in narratives and the role of grammaticised knowledge 

In contrast to French learners of English, German learners of English have to recognise the status accorded to the both types of entities – inanimate elements as well as the protagonist – when selecting information and weighting it for expression in a task of this kind. Table 1. Information selection: Reference to inanimate entities in the learner varieties L2 French–English

L2 German–English

34.1

26.7

average % values for 20 speakers/group

The comparison shows that German–English learners retain the principles found in German, where references to inanimate entities amount to only 24.5 (which is significantly lower compared to English at 34.5), while the figures for French learners of English compares to those observed in English and French. In conclusion, languages in which the syntactic subject plays a prominent role in information structure (that is, in English and the Romance languages through either fixed word order or pro-drop properties), all follow a similar principle and direct attention in information selection to all types of entities. As will be discussed below, inanimate elements are also accorded a higher status in information packaging at clause level (eligibilty for mention as subject of a main clause, for example), as well as in the means used in reference introduction (mentioned as main content of a clause), compared to German, providing other factors do not intervene. As the following section shows, however, French is a case in point where specific factors do intervene.

. Information status at clause level The second level of analysis relates to the informational status accorded to the protagonist and inanimate entities, as reflected in the way information on these two types is coded at clause level. The first question is whether the protagonist and inanimate entities are both eligible for mention as the syntactic subject of a main clause or not: the prominence accorded to inanimate entities in selecting them for mention can be modified and the status ‘subject of a clause’ can be further weighted in information structure by mapping referents as subject of a main or subordinate clause. Packaging options of this kind are basic indicators of informational status and are associated with topic assignment (Foley & Van Valin 1984; Lambrecht 1987, 1994; von Stutterheim 1997; Van Valin & LaPolla 1997). As the cross-linguistic comparison for English, French, German, Italian and Spanish show, there is evidence of intervening variables since there are two determining factors which drive the mapping preferences observed: Eligibility for men-

JB[v.20020404] Prn:26/06/2003; 13:29

F: SIB2610.tex / p.8 (439-481)

 Mary Carroll and Monique Lambert

tion as subject of a main clause (protagonist only, or protagonist + inanimate entities) is linked to the types of temporal concepts which are grammaticised for the present tense in these language (tense, aspect). In the cross-linguistic comparison there are two languages in which the temporal category aspect is marked grammatically on the verb for the present tense, English and Spanish, and two in which tense only is grammaticised for the present, French and German. The present tense plays a central role in the analyses since this was the tense chosen by the majority of the speakers, as mentioned above. In a narrative about the quest of a single protagonist for water in desert-like worlds, reference to the protagonist as subject of a main clause can be expected to occur frequently in the narrative sequence. The question is do the languages studied accord similar status to all entities which potentially qualify for mention as subject of the main clause or not? English and Spanish show similar preferences in the coding options selected at clause level, in that inanimate entities are eligible for mention as subject of a main clause. This is not the case in French and German, where such entities are mainly mapped as subject of a subordinate clause, as in French, or are coded in passive constructions as in German (German: er wird von einem grossen Blatt getroffen, ‘he is knocked over by a large sheet of paper’). The cases counted across the languages relate to utterances which belong to the main structure or narrative sequence (chain of events) and exclude occurrences in subordinate clauses when introducing entities to the domain of discourse (there are large pieces of paper which are flying around), or in comments. Even though English and the Romance languages direct attention to inanimate enitites in information selection, only English and Spanish (+ grammaticised aspect, present tense) assign a similar status to these elements in that they can occur as the subject of a main clause: English 26.0, Spanish 24.1, while the values in French and German are lower with French 10.4 German 14.7 (average % for 20 speakers in each language). In the remaining main clauses there is either the protagonist as subject: English 69.0, French 75.5, German 76.1, or the observer, narrator, whereby the latter references (you get the impression ...) do not vary significantly across the different languages. These mapping patterns for inanimate entities in main clauses are more clearly underlined, however, by the relatively low occurrence of similar references in the role of syntactic subject in subordinate clauses in English and Spanish, compared to French. The occurrence of inanimates as subject of a subordinate clause is relatively low in English at 2.5, Spanish 17.3, and German 12.2, while French shows a marked contrast at 43.1 (average % values for 20 speakers). As the figures indicate, the occurrence of an inanimate entity as subject of a clause is a factor which clearly drives subordination in French. So the two languages with grammaticised aspect for the present tense (English and Spanish) integrate situations involving inanimate

JB[v.20020404] Prn:26/06/2003; 13:29

F: SIB2610.tex / p.9 (481-535)

Information structure in narratives and the role of grammaticised knowledge 

entities into the narrative via a main clause, while these situations generally remain excluded from this sequence in French. Although the findings for German show surface similarities with English in that inanimate entities are not frequently coded in subordinate clauses (the figure is low at 12.2, compared to French), references to inanimates are “demoted” in information structure in German so to speak, compared to the protagonist, since they often occur in passive constructions (er wird von einem Blatt umgehauen, ‘he is knocked down by a sheet of paper’). They are thus not accorded the informational status which allows the speaker to code them as the subject of the clause (the sheet of paper knocks him down) (cf. Murcia-Serra 2001). Furthermore, possible references to these entities are filtered out by selection principles in German in the first place, since speakers relate to such phenomena to a significantly less degree, compared to all the other languages studied. A detailed discussion of these cross-linguistic differences and their link to topic focus assignment goes beyond the scope of the present paper but is given in Carroll, Lambert, & Murcia-Serra, forthcoming. The tendency to assign eligibility for mention in a main clause to animate actors such as the main protagonist in a narrative is given by the fact that their activities in a narrative context typically bring about a change in state. Change of state events are prime candidates in forming a chain of bounded events – the backbone or main structure of the narrative text. In other words, change of state activities shift the narrative along the set time line (first x happens, then y, etc.) and are generally coded in terms of the simple present tense. Aspectual markings, on the other hand, are less likely to be used when a change of state is involved, since the meanings coded in the present languages relate to concepts such as ongoingness, and stand in contrast to bounded events: Example 2. (1) he wakes up (change in state, simple tense) in a strange world which is full of rocks (2) gets up (change in state, simple tense) (3) and takes stock of the situation (change in state, simple tense) (4) he notices (change in state) (5) that one of the rocks is heading straight for him (ongoing situation, marked by the progessive and not a simple tense form, plus subordination) (6) and he just manages to get out of the way (change in state, simple tense) The mapping of ongoing situations into a subordinate clause is more likely to occur when they are mentioned in connection with the narrative sequence or chain of events (and utterance (5) illustrates one of few the cases in which mappings of this kind are found in English). The use of subordinate clauses in this function

JB[v.20020404] Prn:26/06/2003; 13:29

F: SIB2610.tex / p.10 (535-643)

 Mary Carroll and Monique Lambert

in narratives is well documented across a number of languages (cf. Hopper 1979; Tomlin 1986). The cross-linguistic comparisons highlight a significant factor guiding the different mapping patterns observed (main clause vs. subordinate clause). The contrast between change in state and ongoing situations can be easily marked by the contrast simple tense (he jumps) versus the progressive (he is jumping) in English and Spanish. Both temporal structures are differentiated by codings on the verb and thus need not be distinguished via the contrast main and subordinate clause (cf. Hopper 1979). On the surface this looks like a simple situation for the learner – the role accorded to subordination in the one language matches that of the progressive in the other – but analyses of the narratives in English and Spanish show that ongoing events coded in the progressive are often integrated into the temporal sequence of events in ways not found in languages such as French or German. The problem of analysis for French learners, for example, is confounded by the fact that they cannot rely on a clear cut correspondence in function between subordination and the progressive, since certain event types which are frequently coded in the progressive in English are never subordinated in French. And ongoing events are occasionally subordinated, as illustrated in utterance (5) above. The function of the progressive in narrative structure and the integration of different event types within a narrative sequence constitutes one of the core problems for learners of English (von Stutterheim & Lambert, in press; Carroll & von Stutterheim 2002). The cross-linguistic differences in information structure thus pose a number of problems for French and German learners of English. French learners of English have to recognise that activities relating to the protagonist (change in state) and those of inanimate entities can both be integrated into the sequence of events, as syntactic subject of a main clause. Subordination of events involving inanimate entities is not a necessary or dominant feature of information structure in English for sequences of utterances which belong to the main structure of the narrative, as discussed above. Starting the analysis of the learner varieties with main clauses, the results show that French learners of English are less likely to code inanimates as the subject of a main clause, thus maintaining a pattern which compares to their source language. Although German learners have a reduced number of references to inanimate entities in the first place, they have recognised that those selected are eligible for mention in main clauses, in contrast to German. The values found in the narratives of German learners (24.3) compare to those in English native speaker narratives (26.0). The corresponding values for inanimates as subject of a main clause in the other languages are English 26.0, Spanish 24.1, French 10.4, German 14.7. Interestingly, the learner languages both exhibit similar principles by mapping inanimate entities into subordinate clauses to a comparable degree.

JB[v.20020404] Prn:26/06/2003; 13:29

F: SIB2610.tex / p.11 (643-691)

Information structure in narratives and the role of grammaticised knowledge 

Table 2. Inanimate entities as syntactic subject of a main clause in learner varieties

Inanimate entities Protagonist observer/other

L2 French–English

L2 German–English

13.2 76.0 10.8

24.3 67.7 8.0

average % values for 20 speakers/group

Table 3. Subordinate clauses in learner varieties with an inanimate entity as subject

subject: inanimate entity

L2 French–English

L2 German–English

35.1

29.0

average % values for 20 speakers/group

It is useful at this point, however, to relate these values to the role of subordination in general, that is, taking the ratio of all types of subordinate clauses to main clauses, in the different language groups. In English subordinate clauses amount to 19.1, German 11.2, French 37.0, and compare as follows to the learner varieties: Table 4. Ratio of subordinate clauses (to main clauses)

subordinate clauses

L2 French–English

L2 German–English

30.9

9.4

average % values for 20 speakers/group

The use of subordination in the narratives is thus highest in French and L2 French– English, underlining the fact that learners retain the source language principle in this respect also. The values for German narratives and L2 German–English also compare, but the rate of subordination is relatively low in both German, L2 German–English, and English, albeit for different reasons. Some degree of downgrading of inanimate entities, with respect to informational status, is still in evidence in L2 German–English, compared to English, since subordinate clauses frequently have an inanimate as subject (29.0 in L2 German–English in contrast to 2.5 in English). Since use of subordination in this function is not predominant in German or English, this coding pattern is L2 specific. The function of subordination in the learner variety of German learners may be linked to the fact that there is a clear absence of passive constructions in L2 German–English, compared to L1 German narratives, in which inanimate entities form the underlying logical subject of the clause (er wird plötzlich hochgeschoben; he is suddenly pushed up into the air). Subordination may thus constitute an L2specific means of coding similar distinctions in the informational status of inani-

JB[v.20020404] Prn:26/06/2003; 13:29

F: SIB2610.tex / p.12 (691-749)

 Mary Carroll and Monique Lambert

mate entities. But with the explicit mention of an inanimate entity in the role of syntactic subject, albeit in a subordinate clause, the German learners of English have promoted its status from that of an underlying logical subject, which is not explicitly mentioned at all, to one which is mentioned.

. Introduction of inanimate entities and linkage (causal, temporal relations) The final factors studied in uncovering the assignment of informational status relate (a) to the way inanimate entities are introduced into the narrative, comparing how referents are mapped into form and (b) to the types of relations (temporal and causal) used to link utterances in the narrative (linkage).

.. Introduction of inanimate entities The cross-linguistic comparisons of English, French, German, and Spanish show that English and Spanish again follow principles which contrast with those found for French and German. As indicated above, a distinction was drawn between first mentions which occur in their own right and are thus categorised as prominent. prominent: there are sheets of paper swirling around; you can see a chain in the background; or first mentions which are embedded in actions carried out by the protagonist embedded: the figure lands in a new world with lots of paper all over the place; he jumps onto a chain. We assume that more attention is focused on the entity introduced when it involves the main content of the clause, so to speak, while one can say that references are less prominent when embedded in another context (event associated with protagonist). So the categories analysed cover prominent versus embedded introductions. The relevant means which accord prominence to the referent are existentials (there are sheets of paper swirling around), explicit perceptions of elements by the narrator/observer (and then you see a huge rock shooting up out of the ground), explicit perceptions on the part of the protagonist (he sees a chain in the background). These contrast with references embedded in utterances which relate to actions of the protagonist (he lands in a new world which is full of stones; he grabs hold of a chain).The cross-linguistic comparison shows that prominent forms constitute the most frequent means used in English and Spanish. English: prominent 83.1 with embedded at 16.9; Spanish: prominent 70.9 embedded 29.1. Prominent and embedded forms occur with a comparable frequency in French and German. French: prominent 54.6 embedded 45.4, German: prominent 58.6 embedded 41.4. (average % values for 20 speakers/group). In sum, the cross-linguistic analysis shows that the means used to introduce elements correlate with factors determining topic status (tense, aspect).

JB[v.20020404] Prn:26/06/2003; 13:29

F: SIB2610.tex / p.13 (749-822)

Information structure in narratives and the role of grammaticised knowledge 

The comparison with French and German learners of English indicate that they still follow the principles found in the source language, since the difference between the mean values in German and L2 German–English, as well as French and L2 French–English are not statistically significant. Table 5. Reference introduction in learner varieties

Prominent Embedded

L2 French–English

L2 German–English

English

51.9 48.1

67.6 32.4

83.1 16.9

average % values for 20 speakers/group

In summary, French learners of English exhibit L1 influence in the informational status accorded to inanimate entities in their eligibility for mention as subject at clause level (subject of subordinate as opposed to main clause) and the means used in reference introduction (embedded as opposed to prominent), and show a high degree of consistency in the informational status accorded to inanimate entities across the different coding categories studied. German learners of English have recognised principles in the target language which differ from those in their L1, but the learners are not consistent in the informational status accorded to inanimate entities across the different categories. Although there is an increase in the status assigned to such entities in that they are codable as subject of a main clause or subordinate clause, in contrast to passives in German, these entities are nevertheless still demoted, compared to English, taking reference introduction, since they use embedded as opposed to prominent means. Embedded mentions are still high (32.4), compared to English (16.9), where prominent forms predominate (83.1).

.. Linkage and the role of temporal versus clausal relations English, French, Italian, and Spanish all exhibit a clear tendency to link utterances in the narratives via temporal (then, before, later) as well as causal relations (because, and so ...). This is not the case in German where the narratives are marked by the relative absence of causal relations, compared to temporal relations. Table 6. Linkage

temporal causal

English

French

German

61.8 38.1

33.1 66.8

91.1 8.9

average % values for 20 speakers/group

JB[v.20020404] Prn:26/06/2003; 13:29

F: SIB2610.tex / p.14 (822-897)

 Mary Carroll and Monique Lambert

Despite this common tendency to mark causal relations in English and the Romance languages, there are nevertheless differences between English and French in this respect (causal relations English 38.1%, French 66.8%). The prominence of causal relations in French is driven, on the one hand, by the need to create a link between information coded in subordinate clauses, which in many cases involve references to inanimate entities and associated ongoing situations, with the main structure or narrative sequence, as illustrated in the following example. (1) le personnage tombe dans un désert de feuilles de papier ‘the person falls in a desert of sheets of paper’ (2) il y a beaucoup de vent ‘its very windy’ (3) il entend la goutte d’eau ‘he hears a drop of water’ (4) qui tombe ‘which falls’ (5) et donc il se lève ‘and so he gets up’ (6) il marche ‘he walks’ (7) et le vent lui envoie une feuille de papier dans la figure ‘and the wind blows a sheet of paper into the figure’ (8) et il tombe à la renverse ‘and he falls over backwards’ (9) ensuite il se relève ‘then he gets up again’ (10) et il se rend compte ‘and he realises’ (11) que la goutte d’eau tombe sur l’une des feuilles ‘that the drop of water falls on one of the sheets’ (12) et donc il touche cette feuille ‘and so he touches that sheet’ As the example shows, the information linked in via causal relations (donc) often relates to inanimate entities mentioned in the preceding utterance. They correspond with the use of the form so (016) in English in the following excerpt, but unlike French, information on inanimate entities is integrated into the narrative

JB[v.20020404] Prn:26/06/2003; 13:29

F: SIB2610.tex / p.15 (897-963)

Information structure in narratives and the role of grammaticised knowledge 

sequence in main clauses, and use of causal links is not as frequent, as discussed above.2 (1) and there is this rock falling from the sky (2) it is heading straight for him (3) so he quickly rolls over (4) and luckily it lands beside him In contrast, speakers of German focus on change in state situations, which set the stage for temporal relations (und dann, ‘and then’) as the main form of linkage in narratives in German (85.6), while causal links amount to no more than 8.9 (compared to 66.8 in French). Coming now to the learner varieties of English, the results for the German learners show that they have recognised the role of causal over temporal relations in linking utterances in English (causal relations 38.1), compared to the role of causal relations in German (8.9). Table 7. Linkage in learner varieties

temporal causal

L2 French–English

L2 German–English

40.5 59.5

67.7 32.2

average % values for 20 speakers/group

Learners of French have not yet acquired the target language principles, given the increased use of causal relations (compared to English causal 38.1). They still map information on inanimate entities into subordinate clauses, as in French, and thus have not acquired the underlying principles determining use of the progressive tense in narratives, which obviates the necessity of using subordination in English. They therefore have not fullly acquired the functional distinction which is marked on verbs in English narratives between events which are coded in the simple tense (potential change of state) versus the progressive (ongoing). They have acquired the formal means simple tense and progressive, but not the full implications of the functional distinctions which they encode in the narrative sequence.

. Conclusions The empirical findings clearly show where the learner’s problem of analysis lies: in order to acquire the functional role of a distinction such as the simple tense versus the progressive in narratives in English, both groups of learners have to recognise

JB[v.20020404] Prn:26/06/2003; 13:29

F: SIB2610.tex / p.16 (963-1019)

 Mary Carroll and Monique Lambert

its overall status in information structure and with this its role in a clearly defined ‘system of meanings’ or form function relations. The role of this system in information structure starts with the process of information selection or considering ‘what to say’: English and the Romance languages follow common principles in that, in deciding what to say, attention is directed to all types of entities (protagonist plus inanimate entities). The common grammatical feature which English, French, and Spanish share involves the syntactic subject and its prominent role in information structure, which is grammaticised through either fixed word order, as in English and French, or pro-drop features, as in Spanish and Italian. There is tentative evidence that the syntactic subject drives information selection in these languages, since information which maps into the syntactic subject of a clause (entities) is selected for mention without evident restrictions in English and the Romance languages. This contrasts with the selection principles observed in German, where the syntactic subject is grammaticised in a different way. In German the syntactic subject, and the information which it encodes, shares its potential status (encode topic information) with other categories which map into the ‘Vorfeld’ or preverbal ‘slot’ (temporal relations, for example, in narratives). The dual role in information structure in German of both the syntactic subject and constituents which code temporal relations (temporal shift with dann, then), channeled by the verb second constraint, may reduce the scope of the syntactic subject and focus attention in information selection on a specific subset of entities which are associated with certain event types (temporal shift). So the syntactic subject is not tied to a single role in information structure, and shares its clear potential for coding topic information with temporal forms in narratives which alternatively fill the slot preceding the verb (‘Vorfeld’).3 The findings also underline how the mapping option information is eligible for mention as the subject of a clause is systematically modified via the alternatives syntactic subject of a main clause versus syntactic subject of a subordinate clause, and that modification at this level is driven by a further set of grammatical features, tense and aspect (ongoingness). These are both grammaticised on the verb (for the present tense) in English and Spanish, but not in French or German (tense only). In English and Spanish, languages in which the aspectual meaning ongoingness is grammaticised on the verb, both inanimate forces and references to the protagonist are eligible for mention as subject of a main clause. This can be attributed to the fact that different event types (change of state event/protagonist and ongoing events/inanimate entities), a distinction which is relevant in a narrative context, can be marked by the contrast simple tense versus ongoing/progressive aspect in English and Spanish. This is not the case in French, in which progressive aspect is not grammaticised on the verb, and inanimate entities, which often involve ongoing events (sheets of paper are swirling around), are generally mapped as subject of a subordinate clause in order to mark their temporal status within a narrative

JB[v.20020404] Prn:26/06/2003; 13:29

F: SIB2610.tex / p.17 (1019-1074)

Information structure in narratives and the role of grammaticised knowledge 

sequence. Furthermore, the assignment of topic status to entities such as the protagonist and inanimate forces differs in English and Spanish (+ aspect), on the one hand, and French, German (– aspect). Taking another step in the specification of information structure, the informational status reflected in these coding options at clause level (codable as syntactic subject of a main clause or not) correlates with the prominence assigned to entities when introducing them as referents to the domain of discourse (main content of a clause vs. embedded in clauses relating to actions of the protagonist). Similarly, the extent to which speakers link utterances in the narrative by means of temporal or causal relations correlates with factors guiding information selection, driven by the role of the syntactic subject. All languages which are subject prominent in information structure and select information on both the protagonist and inanimate forces (English, French, Italian, Spanish) link utterances by means of both temporal and causal relations. The scope for causal relations is reduced in German since change in state events and associated temporal links (temporal shift) predominate in the narratives. There is thus evidence of clusters of form function relations which are typolgically relevant and correspond to those found in studies of descriptive texts. As discussed in detail in Carroll, Lambert, & Murcia-Serra, forthcoming, these mapping preferences illustrate clusters or sets of associated principles in information structure which come to the fore in the context of the present analyses. There is a hierarchy of coding options starting with syntactic subject of a clause – syntactic subject of a main clause > syntactic subject of a subordinate clause > not eligible as syntactic subject of a clause but as underlying logical subject (inanimate forces in the role of instrument in passives: he is pushed up by a huge rock). The analysis of the learner varieties provide relevant insights into the nature of the factors guiding information structure. None of the learners (in this case L2 German–English) have revised or reorganised the role of the core factors which guide the type of information which is selected for mention at the outset of the task. In deciding what to say at the level of macro-planning, German learners of English do not direct attention, for example, to inanimate entities to the degree found in the ‘subject prominent’ languages in the group, even though they are beginning to recognise that the inanimate elements they select are (a) eligible for mention in main clauses, and that (b) causal relations are often used in relating to circumstances surrounding actions of the protagonist (often ongoing states such as water is dripping) and in linking them on a causal basis into the narrative sequence. However, they have not yet accorded the required status to inanimate entities in the means used in reference introduction, given the predominance of embedded over prominent forms of introduction in their narratives. So there is still evidence of downgrading or demotion, so to speak, in the means used in reference introduction, and in the tendency to map inanimates as subject of a subordinate

JB[v.20020404] Prn:26/06/2003; 13:29

F: SIB2610.tex / p.18 (1074-1136)

 Mary Carroll and Monique Lambert

clause, although there is nevertheless a clear departure from the status mappable as instrument in a passive construction, as typical in German. Further studies are required in this context of more advanced or near native speakers of English as a second language in order to see how far their analyses proceed in uncovering the principles of the target language English. Very advanced Spanish learners of German (cf. Murcia-Serra, this volume) do revise the source language principle, whereby agentivity and not topic determines whether an entity becomes subject of a main clause or not, and clearly acquire the target language pattern where the entity which is topic maps into the subject of a clause, irrespective of its role (agent or undergoer). It may be easier in acquisition to track a clear cut restriction to one mapping option (topic only), compared to the task facing French learners who have to recognise the role of aspectual markings in coding a core distinction in the narrative (between ongoing versus bounded or change of state events), and link this to the syntactic subject, thus obviating the use of subordination in this function. Furthermore, the Spanish learners of German (same group) are not entirely successful in all respects since they do not acquire the principles underlying information structure in descriptive tasks in German but maintain very basic constraints of their source language (Carroll, Murcia-Serra, Watorek, & Bendiscioli 2000). On the whole French learners of English have a better headstart it would seem since some of the core principles found in the selection of information in the narrative task coincide in English and French. But, as mentioned above, they have not acquired use of the functions which aspectual markings can serve in narrative structure in English. Since the status accorded to tense and bounded events also drives topic assignment, as in French, they retain associated principles of the source language in the means used both in introducing entities to the domain of discourse and mapping them as subject of a main clause within the narrative sequence, since these form an entire cluster in both learner and fully fledged varieties. The overall consistency or clusters found in information structure in L2 French–English may prove an impediment to further reorganisation, while the absence of consistency in L2 German–English may provide futher impetus for change. It is still an open question, however, as to whether adult learners can fully succeed in unravelling the “system of meanings” or form function relations which hold in information structure in a given language. The cross-linguistic analyses give an indication of the nature of the learner’s task and the difficulties they encounter in uncovering the role assigned to only a specific set of grammaticised features of a language which work in unison to promote coherence in information structure. When acquiring a first language learners can start with the grammatical building blocks and slowly discover their relevance in information structure as development proceeds. But as adult speakers they know that the specification of information structure proceeds in principled terms from the very outset in macro-planning,

JB[v.20020404] Prn:26/06/2003; 13:29

F: SIB2610.tex / p.19 (1136-1182)

Information structure in narratives and the role of grammaticised knowledge 

and there is no evidence in the studies conducted so far that knowledge of the abstract principles involved (assign topic status, for example) is fully set aside when approaching the task of acquiring a second language (cf. Slobin 1993). In the early stages (basic variety) learners follow principles which are L2 specific and use over explicit forms (full NPs as opposed to pronouns) in anaphoric linkage, for example (cf. Gullberg, this volume). As proficiency increases and texts increase in complexity, speakers start to adapt L2-specific principles to those of the target language. In doing so, however, source language principles also seem to become eligible – as working hypotheses at least. So the findings provide a window on basic questions relating to the status of the linguistic knowledge on which adult learners of a second language proceed (see Klein 1996; Birdsong 1999). Unlike first language acquisition, adult learners do not go back to a base line and proceed from there. As the present findings for advanced learners indicate, information on the linguistic system of the second language is filtered by abstract principles which are required in information structure, principles which are themselves linked to sets of grammaticised meanings in the learner’s source language. Advanced learners have to dissociate universal principles in information organisation such as “assign topic status” from the set of grammatical means which determine the choices made in their L1, and uncover the set of grammaticised forms which determine the selections made in the target language. The relevance of grammaticised systems of meanings in information structure pointedly illustrates why native-speaker narratives sound native-like and those of second language learners, though formally correct, do not. The barriers to nearnative competence are not cultural but grammatical in nature, and the role of the relevant system of meanings and the clusters they form is proving to be an almost insurmountable problem of analysis in second language acquisition.

Notes . The expletive subject in the existential there is, is a stand in, so to speak, for the topic ‘picture’, and the statement there is an x could be paraphrased as the picture has . . . . . Mapping patterns found in Italian correspond to those in French, cf. Carroll, Lambert, Murcia-Serra, forthcoming. . A detailed analysis of the factors determining information structure in German go beyond the scope of the present paper and is given in Carroll, Lambert, Murcia-Serra, forthcoming.

JB[v.20020404] Prn:26/06/2003; 13:29

F: SIB2610.tex / p.20 (1182-1272)

 Mary Carroll and Monique Lambert

References Birdsong, D. (Ed.). (1999). Second language acquisition and the critical period hypothesis. Mahwah, NJ: Erlbaum. Berman, R. & Slobin, D. I. (1994). Relating events in narratives: A Crosslinguistic Developmental Study. Hillsdale, NJ: Erlbaum. Carroll, M., Murcia-Serra, J., Watorek, M., & Bendiscioli, A. (2000). The relevance of information organisation to second language acquisition studies: the descriptive discourse of advanced adult learners of German. Studies in Second Language Acquisition (Special Issue ed. by Clive Perdue), 22(3), 441–466. Carroll, M., Lambert, M., & Murcia-Serra, J. (forthcoming). Information structure and grammaticalisation: a cross-linguistic comparison of Arabic, English, German, and Romance languages. Manuscript: University of Heidelberg. Carroll, M. & Stutterheim, C. von (2002). Typology and information organisation: the question of perspective taking and language-specific effects in the construal of events. In A. Ramat (Ed.), Typology and Second Language Acquisition. Berlin: de Gruyter. Foley, W. A. & Valin, R. van (1984). Functional syntax and universal grammar. Cambridge: Cambridge University Press. Givón, T. (Ed.). (1983). Topic Continuity in Discourse: A Quantitative Cross-language Study. Amsterdam: Benjamins. Hockett, Ch. (1954). Chinese versus English: an exploration of the Whorfian thesis. In H. Hoijer (Ed.), Language in Culture (pp. 106–123). Chicago: University of Chicago Press. Hopper, P. (1979). Aspect and foregrounding in discourse. In T. Givón (Ed.), Discourse and Syntax (pp. 213–241). New York: Academic Press. Klein, W. (1996). Language acquisition at different ages. In D. Magnusson (Ed.), The lifespan of development of individuals: behavioral, neurobiological, and psychosocial perspectives: a synthesis (pp. 244–264). Cambridge: Cambridge University Press. Lambrecht, K. (1987). Sentence focus, information structure, and the thetic-categorical distinction. Proceedings of the Thirteenth Annual Meeting of the Berkeley Linguistics Society, 366–82. Berkeley, CA. Lambrecht, K. (1994). Information Structure and Sentence Form. Cambridge: Cambridge University Press. Levelt, W. J. M. (1996). Perspective taking and ellipsis in spatial descriptions. In P. Bloom, M. A. Peterson, L. Nadel, & M. F. Garrett (Eds.), Language and Space (pp. 77–107). Cambridge, MA: The MIT Press. Levelt, W. J. M. (1999). Producing spoken language: a blueprint of the speaker. In M. C. Brown & P. Hagoort (Eds.), The Neurocognition of Language (pp. 83–122). Oxford: Oxford University Press. Murcia-Serra, J. (2001). Grammatische Relationen im Deutschen und Spanischen. Frankfurt am Main: Peter Lang. Slobin, D. I. (1991). Learning to Think for Speaking: Native Language, Cognition, and Rhetorical Style. Pragmatics, 1(1), 7–25. Slobin, D.I. (1993). Adult language acquisition: a view from child language study. In C. Perdue (Ed.), Adult language acquisition: cross-linguistic perspectives (pp. 239–249). Cambridge: Cambridge University Press.

JB[v.20020404] Prn:26/06/2003; 13:29

F: SIB2610.tex / p.21 (1272-1309)

Information structure in narratives and the role of grammaticised knowledge 

Slobin, D. I. (1996). From ‘thought and language’ to ‘thinking for speaking’. In J. J. Gumperz & S. C. Levinson (Eds.), Rethinking linguistic relativity (pp. 70–96). Cambridge: Cambridge University Press. Stutterheim, C. von (1997). Einige Prinzipien des Textaufbaus. Tübingen: Niemeyer. Stutterheim, C. von & Lambert, M. (in press). Crosslinguistic analysis of temporal perspective. In H. Hendricks (Ed.), The structure of learner varieties. Berlin: de Gruyter. Stutterheim, C. von & Nuese, R. (in press). Processes of conceptualisation in language production. Linguistics (Special Issue: Processes of perspectivation in language production). Talmy, L. (1987). The relation of grammar to cognition. In B. Rudzka-Ostyn (Ed.), Topics in Cognitive Linguistics (pp. 165–205). Amsterdam: John Benjamins. Tomlin, R. S. (1986). Foreground-background information and the syntax of subordination. Text, 5, 85–122. Wierzbicka, A. (1995). A semantic basis for grammatical typology. In W. Abraham, T. Givon & S. A. Thompson (Eds.), Discourse Grammar and Typology (pp. 179–209). Amsterdam: John Benjamins. Valin, R. D. van & LaPolla, R. (1997). Syntax. Cambridge: Cambridge University Press.

JB[v.20020404] Prn:26/06/2003; 13:36

F: SIB2611.tex / p.1 (35-100)

Acquiring the linkage between syntactic, semantic and informational roles in narratives by Spanish learners of German* Jorge Murcia-Serra

.

Introduction

The syntactic relation subject is in general considered to be the target coding for information which for different conceptual factors may be regarded by the speaker to be the most salient in the utterance (cf. Fillmore 1977; Turner & Rommetveit 1968; Levelt 1989). The factors responsible for salience mentioned in the different studies on the category subject can be roughly reduced to three: 1. The focus of attention of the speaker: entities which are within the centre of interest of the speaker are more salient than those which are not (cf. J. M. Carroll 1958; Kuno & Kaburaki 1977; Kuno 1983; Tannenbaum & William 1968; Turner & Rommetveit 1968; Tomlin 1995, 1997). 2. The inherent or perceptual properties of the entities referred to in speech such as size, movement, animacy, causation, or concreteness in the sense that entities which (in relation to others) are big, animate, in movement, causing the action depicted and/or concrete are perceived by the speaker to be more salient (cf. Bock & Warren 1985; Bock et al. 1992; Flores d’Arcais 1975; Osgood & Bock 1977; Sridhar 1988; see also the distinction “Figure/Ground” by Talmy 1978, 1988). 3. The relative importance of determined entities in a complex state of affairs, as shown by the frequency or continuity of reference to that entity in a given text (cf. Givón 1983, 1992, 1994). This is a consequence of the unfolding of information in discourse in accordance to the communicative intention of the speaker (cf. Levelt 1989; Stutterheim 1997). In most cases, these three factors are interrelated in such a way that on the whole they can be regarded to correspond with two linguistic categorisations: topic and

JB[v.20020404] Prn:26/06/2003; 13:36

F: SIB2611.tex / p.2 (100-149)

 Jorge Murcia-Serra

agent (cf. Bock & Levelt 1994). The category topic roughly accords with the reference to that entity, for which the assertion formulated by the corresponding utterance holds (cf. Reinhart 1982; Vallduví 1992, 1993; Primus 1993; Lambrecht 1994). The thematic role of agent, on the other side, can be considered to possess prototypical and perceptually relevant inherent features like intentionality, movement or causation (cf. Dowty 1991; Van Valin & La Polla 1997), which in turn call for the attention of the speaker and might increase the importance of a determined entity in discourse. This interrelation leads in many languages (in the case of the so-called accusative languages) to the linking of the three categories: the entity regarded by the speaker to be the topic in a given state of affairs has at same time the control over that state of affairs will be coded as the subject of the corresponding utterance as shown in (1).1 (1) A: What did Brian do? B: He took a cross and walked up the hill A possible conflict for this prototypical linking arises especially in discourse production whenever a topic entity is implied as a patient in a given state of affairs. In this case, languages may use different means to track for reference to both topic and agent (cf. Van Valin & La Polla 1997). Some languages, with a relatively fixed constituent order, as for example English, rely basically on passive constructions to show the parting of reference of topic and agent, as can be seen in (2). (2) A: What about Calvin? B: He was eaten up by a green alien Thanks to the use of a passive, the topic (patient) entity can be coded as the subject of the utterance in an initial (topical) position while the agent entity (if necessary) may appear as a prepositional phrase in a later (focal) position (cf. Primus 1993). In languages with a more flexible constituent order, a topic entity is basically coded by exploiting such a flexibility, typically through left-detachment, independently of whether it concerns a subject or an object constituent, an agent or a patient argument. Due to the affinity between subject and agent, which has been repeatedly observed experimentally (cf. for instance Bock & Warren 1985; Igoa 1996) a non-topical agent entity will be coded in such languages as a subject in a post-verbal (focal) position. This is the case for example in Catalan (Romance) as showed in (3) (cf. Vallduví 1992, 1993).2 (3) A: Què hi ha d’en Josep? What there has of the Josep? ‘What’s new about Joseph?’

JB[v.20020404] Prn:26/06/2003; 13:36

F: SIB2611.tex / p.3 (149-203)

Acquiring the linkage between syntactic, semantic and informational roles 

B: En Josep el va atropellar ahir un cotxe The Joseph him goes hit-inf yesterday a car ‘Josep was hit by a car yesterday’ A derived consequence of these differences between languages in relation to reference tracking systems is the function that the category subject assumes in discourse production. In languages that consistently match the topic entity in the sentence with the subject category (like we have seen for English), this category assumes an important role (especially in the case of narratives) in establishing cohesion on a global text level since it acts as an anchor point for reference to a global topic entity, which is typically the protagonist of the narration being produced. In languages which map agent-subject (like Catalan), the subject category does not act as an anchoring for textual cohesion the close affinity between both categories leads in some cases (as we will see in detail) to the setting of a local topic (for one or a few utterances), which is to be distinguished from the global topic at a text level (see next section), although they may both coincide. Topicality is marked in these languages via constituent order as opposed to a specific grammatical category, but the subject category still maintains a privileged function for the coding of topical information since it can be placed in a topical position in the utterance and allows minimal reference by using zero anaphora (especially in the case of pro-drop languages thanks to agreement marking on the verb). For an adult native speaker of a language which maps agent–subject and uses constituent order to track for reference to the topic (the case of Catalan), the task of analysing and learning a language matching topic with the subject category (like in the English examples we have seen) poses a number of problems: it is necessary on the one hand to recognise the relevance in the target language of the category subject for textual cohesion, and on the other hand to reduce the outstanding role of agentivity for salience, and so to avoid the setting of a local topic, which in the target system would mean a break in textual cohesion. To study the nature of such an acquisition process, we will examine data from learners of German as a second language with Spanish as their native tongue. As we will see, German favours passive constructions to secure the tracking of the global topic entity in narratives (i.e. the protagonist) irrespective of its thematic role (agent or patient), while Spanish (like Catalan) basically relies on constituent order to track the topic entity, thus giving greater prominence to agentivity for coding as the subject.

JB[v.20020404] Prn:26/06/2003; 13:36

F: SIB2611.tex / p.4 (203-258)

 Jorge Murcia-Serra

. Subject, topic and agent When dealing with categories such as subject, topic and agent for which there are different positions as for their definition, it is necessary to state the way in which they are to be understood. The syntactic relation subject is defined in the present paper in formal terms as the constituent showing agreement with the finite verb and, when relevant, marked with the nominative case. Besides, in Spanish, being this a so-called pro-drop language, the subject is the only core constituent that can be deleted from the sentence without requiring any kind of overt trace as a weak pronoun (cf. Fernández Lagunilla & Anula Rebollo 1995). Such a definition, which in terms of typological universality is for sure more than problematic (cf. among others Keenan 1976) should be appropriate for the languages at issue here, i. e. German and Spanish.3 As for semantic roles, we will use here the categories Actor and Undergoer in accordance with Van Valin and LaPolla (1997).4 These categories are conceived as macroroles which represent generalisations for the argument-types prototypically represented by the traditional roles of agent and patient respectively. In this sense, the macrorole Actor stands for roles like agent, experiencer, instrument, force, or possessor. Undergoer may encompass roles like patient, theme, recipient, location, or possessed.5 Topic is to be understood as the reference to the entity about which the speaker wants to give new information to the addressee or in logical terms, the entity for which the assertion contained by the corresponding proposition holds (cf. among others Reinhart 1982; Vallduví 1992, 1993; Primus 1993; Lambrecht 1994). In a communicative sense, the formulation of a topic gives the speaker the possibility to make clear for the addressee to what entity the offered information should be mapped to (cf. Givón 1992; Heim 1982; Vallduví 1992, 1993). Furthermore, we distinguish between a local and a global topic. When producing a narration, speakers concentrate in most of the cases on the actions and events concerning one single entity, typically a person, who is regarded to be the protagonist of the narration and is therefore in the focus of attention throughout the text. The protagonist is thus a primary candidate to be treated as a topic in the sense stated above. This we will call global topic (cf. the concept of main topic in Givón 1983, 1992, 1994). Nevertheless, it is always possible that in the course of her narration the speaker switches her attention to some other entity implied in the state of affairs at issue, typically as a consequence of the kind of inherent perceptual features named above. In this way, such an entity can be treated as a topic in the sense used here for one of a few utterances. For this we use the term local topic. The difference in status (as primary candidate for topicality) can be seen for example in the way in which a re-introduction of reference to a global topic takes place in comparison to the re-introduction of a local topic: a global topic requires

JB[v.20020404] Prn:26/06/2003; 13:36

F: SIB2611.tex / p.5 (258-337)

Acquiring the linkage between syntactic, semantic and informational roles 

typically a minimal form for re-introduction like an unstressed pronoun or in the case of null-subject languages merely a zero-anaphora while the re-introduction of a local topic on the rule takes place by using a full-NP, often accompanied by a relative clause.

. Data The study is based on data collected from 15 speakers of German as a second language with Spanish as their native language. The speakers of this group showed some differences in their proficiency of the German language but in general they can be considered to be advanced to very advanced. They all had studied or were still studying at the University of Heidelberg and all of them had had previous formal teaching in the language before studying in Germany. Their time of stay in the country ranged between 5 and 14 years at the time of the data collection. According to different grammatical features in relation to their performance of the target language (constituent order – especially the position of the finite Verb, case marking, null-subject, vocabulary), speakers were divided into three groups. The description of the German and Spanish systems presented here is based on the data of 20 native Speakers per language and is part of a wider study on grammatical relations and patterns of conceptualisation (cf. Murcia-Serra 2001). The task consisted of the re-telling of the animation silent Film “Quest” (Stellmach & Montgomery 1996)6 where the protagonist (a human figure made out of clay) is confronted in his search for water with a series of difficulties which bring him repeatedly in the role of an Undergoer. The film was presented once in full length (about 10 min.) and then a second time with an interruption after every change of scenery (5 on the whole). During this interruptions speakers were asked to re-tell the events of the last scene showed. For illustration of some points of the German and Spanish native systems, some examples from a second set of data on film re-telling (Modern Times) will be used.

. German and Spanish As showed in Murcia-Serra (2001), the subject category in German is basically connected with the primary topical entity globally defined for the text as a whole (i.e. in narratives, with the protagonist). Consequently, when the global topic entity happens to play the role of an Undergoer in a determined state of affairs, the parting of topic and Actor is coded in German by the use of passive constructions so

JB[v.20020404] Prn:26/06/2003; 13:36

F: SIB2611.tex / p.6 (337-394)

 Jorge Murcia-Serra

that the Actor entity (if mention is required) appears in a demoted position. This is shown in example (4): (4) 1

2

3

4

5

6

7

8

9

und ja dann denkt er sich was anderes aus and yes then thinks he himself something different out ‘and yes then he comes up with something different’ geht in so ein Café rein goes in sort-of a café there-in ‘he goes into a sort of café’ lädt sich da auf sein Tablett Berge von Essen loads himself there onto his tablet mountains of food ‘loads his tablet with mountains of food’ ißt dann auch alles eats then also all ‘then he also eats it all’ geht dann zur Kasse goes then to the cash- register ‘then he goes to the cash register’ legt seinen Kassenzettel vor puts his sales-check forward ‘presents his sales check ’ und kann dann natürlich nicht bezahlen and can then of-course not pay ‘and then, of course, he cannot pay’ ruft vorher auch gleich einen Polizisten her calls before also right away a policeman hither ‘before that he also calls out a policeman right away’ und dann wird er von diesem Polizisten in so einen Polizeiwagen gebracht and then gets he by that policeman into sort-of a police-car brought ‘and then he is brought by the policeman to a sort of police car’

As can be seen, in all utterances the category subject is used to refer to the protagonist, regardless of whether this entity carries the role of an Actor or an Undergoer. Furthermore, as utterances no. 1, 8 and 9 in ex. (4) show, in German the formal means used in coding a global topic entity is not constituent order, but rather the subject category. The subject thus receives an important function as carrier of cohesion for the whole text, since it is used to track the primary topical entity. This must be regarded as a consequence of text planning at a global level as a result of the overall communicative intention of the speaker (cf. Stutterheim 1997).

JB[v.20020404] Prn:26/06/2003; 13:36

F: SIB2611.tex / p.7 (394-418)

Acquiring the linkage between syntactic, semantic and informational roles 

In Spanish by contrast, a topic entity is tracked in the same situation by means of constituent order (i. e. utterance initial position), while the Actor entity is coded as the subject of the utterance as the following example shows: (5) 1

y el Chaplin and the Chaplin ‘and Chaplin’ 2 como quiere irse a la cárcel since wants to go-himself to the prison ‘since he wants to go to prison’ 3 se va a una cafetería himself goes to a café ‘goes into a café’ 4 se llena dos bandejas de comida himself fills two tablets with food ‘loads two tablets with food’ 5 y empieza a comer ahí un montón and starts to eat there a lot ‘and begins eating quite a lot’ 6 y cuando va a pagar and when goes to pay ‘and when he goes to pay’ 7 pues pica al escaparate so knocks on the window ‘he knocks on the window’ 8 que en la calle hay un policía that in the street (there) is a policeman ‘because there is a policeman on the street outside’ 9 entra el policía enters the policeman ‘the policeman comes in’ 10 y le dice al policía and him says to-the policeman and he says to the policeman 11 pague pay ‘you pay’ 12 y entonces claro lo detienen Ø and then of course him arrest (3 Pers. Pl.) ‘and then, of course, they arrest him’

JB[v.20020404] Prn:26/06/2003; 13:36

F: SIB2611.tex / p.8 (418-462)

 Jorge Murcia-Serra

In utterances no. 1 to 7 the subject refers to the protagonist. But in all these cases, this entity plays the role of an Actor in the events being presented. In utterance no. 9 the subject (‘el policía’ – reference maintenance in relation to utterance no. 8) makes reference to an entity playing the role of an Actor, but which does not correspond with the global topic. This utterance is consequently formulated as topicless (or thetic, cf. Sasse 1987; Ulrich 1985) as shown formally by the post-verbal position of the subject (cf. González de Sarralde 2000; Neumann-Holzschuh 1997). In utterance no. 10 the subject category refers back to the global topic entity (reference re-introduction, zero anaphora) and this reference is coded as subject, since the protagonist plays here again the role of an Actor. In utterance no. 12, again, the global topic entity is not the Actor of the depicted event, so it appears as a clitic in a pre-verbal position while the subject once more refers to an Actor. But in this utterance the subject corresponds with what is called arbitrary plural: the subject pronoun is dropped (zero anaphora) and the verb shows agreement for the 3rd Pers. Pl. The event depicted by the utterance is to be interpreted as being carried out by an unspecific person, similar to the English form someone (cf. Jaeggli 1986; Suñer 1983). Thus, the entity referred to by the subject here, not being referential, cannot be topic of the utterance (cf. Reinhart 1982). The figures in Table 1 (adapted from Murcia-Serra 2001) show the use of the category subject in German and Spanish in the re-telling data of the film Modern Times as a coding for the protagonist of a narration (global topic) on the one side and for Actor entities on the other side all of them in states of affairs in which the protagonist finds himself in the position of an Undergoer. The mismatch between the percentages for global topic and for Actor comes about due to the production of utterances, in which the protagonist, although affected by the action of a second entity is presented by the speaker as an Actor, for example: “he goes out of prison” as opposed to “he is released from prison”. As it can be seen in Table 1, the subject category in Spanish is always used to refer to the Actor entity. That is, passive constructions are not found in the Spanish data. Consequently, the subject cannot be considered in Spanish to give a line for textual cohesion. In German in contrast, most subjects refer to the protagonist, so giving a line for textual cohesion. Still, speakers of Spanish do keep track

Table 1. Use of subject as coding for Actor and global topic where the protagonist is Undergoer

German Spanish p

Actor

Global topic

14.9 % 100 % < .001

88.1 % 21.6 % < .001

JB[v.20020404] Prn:26/06/2003; 13:36

F: SIB2611.tex / p.9 (462-514)

Acquiring the linkage between syntactic, semantic and informational roles 

of the global topic entity as shown by through constituent order regularities: the protagonist, coded as a clitic, appears in initial position. This picture changes though as soon as the entity causing the event depicted happens to be inanimate. In such a case, a reference to an unspecific Actor entity through an arbitrary Subject (subject pronoun drop and verbal agreement for the 3rd Pers. Pl.) is not possible, since as we have seen, this kind of reference always requires a human unspecific Actor (someone). This fact, combined with the prototypical linking of subject-topic-Actor leads in Spanish to the setting of a local topic. This can be seen in example (6) from the Quest data: (6) 1

2

3

4

5

6

7

8

9

la figura humana de barro the figure human of clay ‘the human clay figure’ que cae del cielo en ese paisaje that falls from-the sky in this landscape ‘that falls from the sky into this landscape’ según se levanta while himself stands up ‘as he is standing up’ tiene que esquivar una lámina de papel has to avoid a sheet of paper ‘has to avoid a sheet of paper’ que viene por el aire empujada por el viento that comes through the air pushed by the wind ‘that flies through the air pushed by the wind’ se incorpora himself straightens up ‘he straightens up’ esquiva unas cuantas láminas avoids some few sheets ‘avoids some sheets’ y a lo lejos se ve un torbellino formado por papeles también and at the distance itself sees a twister (= SUBJECT!) formed by papers too ‘and at the distance it can be seen a twister formed by sheets of papers too’ debe de ser muy grande must be very big ‘it must be very big’

JB[v.20020404] Prn:26/06/2003; 13:36

F: SIB2611.tex / p.10 (514-574)

 Jorge Murcia-Serra

10 porque parece que está muy lejos because seems that is very far-away ‘because it seems to be far away’ 11 y en un momento se gira and in a moment himself turns-round ‘and at some point he turns round’ 12 y una de esas láminas grandes le da en el cuerpo y lo tira and one of those sheets big him hits on the body and him knocksdown ‘and one of those big sheets hits him on the body and knocks him down’ In utterances no. 1 to 7 the subject refers to the global topic entity which at the same time carries the Actor role. Utterances no. 8 to 10 give information on the scenery. The entity coded as subject (torbellino ‘twister’) carries here the role of a theme in a series of state predicates. Since it is not an Actor it is not treated as a topic and consequently it appears in utterance no. 8 in post-verbal position. The subject of utterance no. 11 refers back to the global topic entity (the clay figure), which again accords with the role of an Actor. Specially interesting is utterance no. 12. In this case the subject of the utterance appearing in a topical pre-verbal position refers to the entity carrying the role of an Actor (paper) and is not the global topic entity. Since this entity is not human, the use of an arbitrary plural subject is not possible. An alternative to this would be to place the subject in a post-verbal position and so to present it as non-topical (le da en el cuerpo una de esas láminas ‘him hits on the body one of those sheets’). But this alternative is rarely found in the data (2/20). This means, that speakers of Spanish clearly prefer to choose an inanimate entity to be the topic (locally) of the utterance (even to the disadvantage of the global topic entity), whenever this entity plays the role of an Actor. Consequently, the subject category can be said to be used to track for agentivity. This is in clear opposition to what we observe in German, as example (7) for the same scene shows: (7) 1

2

3

und die Welt die besteht aus so einer Ebene and the world it is made out sort-of a plateau ‘and the world is made consists of something like a plateau’ aber die ist voll mit so Pappkartons but it is full with sort-of card-boards ‘but it is full of something like card boards’ und er fällt da hin vom Himmel and he falls thither from-the sky ‘and he falls in there from the sky’

JB[v.20020404] Prn:26/06/2003; 13:36

F: SIB2611.tex / p.11 (574-624)

Acquiring the linkage between syntactic, semantic and informational roles 

4

bleibt dann erstmal auf dem Rücken liegen remains then at-first on the back lying ‘remains at first lying on his back’ 5 steht dann auf stands then up ‘then he stands up’ 6 und guckt um sich and looks around himself ‘and looks around’ 7 und läuft dann ziellos durch diese Gegend and walks then aimless through this area ‘and then he walks aimless trough that area’ 8 und überall fliegen diese Pappkartons da rum and everywhere fly those card boards there around ‘and everywhere, there are those card boards flying around’ 9 und irgendwann fliegt ihm so ein Pappkarton ins Gesicht and at-some-point flies him one such cardboard onto-the face ‘and at some point, one of the card board flies onto his face’ 10 und dann wird er umgeschmissen and then gets he knocked down ‘and then he gets knocked down’ As can be seen, in the case of inanimate Actor entities, the cohesive function of the category subject in referring to the global topic entity is parallel to what we saw for cases in which the Actor entity happens to be human. That is, German shows no difference in the use of the subject category in dependence of the animacy of the Actor entity. Table 2 shows the results obtained for Spanish and German for the use of the subject category as a coding for Actor and for the protagonist (global topic) again in such states of affairs in which the protagonist plays the role of an Undergoer. The data analysed here is the re-telling of the film “Quest”, where the role of an Actor is played by inanimate entities as long as the protagonist is not active himself.

Table 2. Use of subject as coding for inanimate Actor and for global topic where the protagonist is Undergoer

German Spanish p

Actor

Global topic

26.7 % 81.3 % < .001

73.3 % 20.8 % < .001

JB[v.20020404] Prn:26/06/2003; 13:36

F: SIB2611.tex / p.12 (624-670)

 Jorge Murcia-Serra

As the results show, in comparison with states of affairs having an animate Actor (cf. Table 1) Spanish shows a decrease in the tendency to code Actor entities as the subject of the corresponding utterances. This corresponds with some reduced use of passive constructions in this context which was not observed in the case of animate Actor entities. Consequently it can be said that although the passive certainly poses as a coding option in Spanish (certainly often to be found in the written language), it is not found in our data of spoken speech for such state of affairs in which the non-topical Actor happens to be human. As our data shows, the use of passive in the spoken language is confined to states of affairs in which the global topic entity accords with the role of Undergoer while the Actor role is played by an inanimate entity. Still, the use of passive in such a context is not the preferred form (cf. Table 2). Native speakers of Spanish prefer the setting of an inanimate Actor entity as topic of the respective utterances, demoting the reference to the protagonist in a non-topical position. In German this is not the case, since speakers of this language basically keep to the linking of subject and global topic whatever animacy features the Actor entity might have. In consequence, it can be said that the differences observed are not just a question of the use of a different coding system for the same content (subject coding vs. initial utterance position), but actually a question of what is regarded to be the topic for which the asserted information holds. In other words, we are confronted here with differences of conceptualisations for the same state of affairs. While speakers of German are producing assertions relating to the protagonist (“dann wird er umgeschmissen”) speakers of Spanish produce utterances about the acting entity in the same state of affairs (“una de las láminas lo tira”). We assume that such differences of conceptualisation do not correspond with differences in the way speakers perceive things in the world as a consequence of the structure of the different languages (in the deterministic sense of Whorf 1956), but rather with differences in the way in which concepts are organised at the moment of speaking, as a consequence of the systematic categorisations and structures of the diverse languages. This is what has been termed by Slobin (1991, 1996) “thinking for speaking”. The idea behind Slobin’s concept is that children, when acquiring their first language, learn to pay special attention to specific concepts and relationships as a result of the structural patterns of the language they are acquiring. The relevant structural differences between German and Spanish are the following: German is well known for being a so-called V2-language. This means that the second constituent in the utterance must be on the rule the finite verb. Non finite verbal elements appear in German in final position. Otherwise, German shows a relatively free constituent order. Nevertheless, subject and finite verb are bound by a subjacency rule, i. e. the subject in German must be placed either in the one position before the verb or right after it. This, together with the fact that subjects are

JB[v.20020404] Prn:26/06/2003; 13:36

F: SIB2611.tex / p.13 (670-720)

Acquiring the linkage between syntactic, semantic and informational roles 

optimal for reference maintenance and thus for reference to a global topic (since they are often involved in co-ordination ellipsis), makes the subject category in German an ideal coding for reference to the protagonist of a narrative, thus providing cohesion in discourse overtly. In this way, children acquiring German as a first language learn to pay special attention to the global topic entity of a series of events. Spanish, on the other hand, does not shows any restrictions for constituent order in general and for the position of the subject in particular (although there are clear regularities, cf. Neumann-Holzschuh 1997; González de Sarralde 2000). At the same time, it is a null-subject language, that is, a language which thanks to the richness of its verbal agreement marking system, allows (or rather requires) subjects to be deleted, whenever the identification of the referent is pragmatically or contextually warranted. However, in order to facilitate reference identification for the addressee it is suitable to accord systematically the subject category with Actor entities, since this category is not obligatorily restricted in Spanish to a topical position in the utterance. As a consequence, children acquiring Spanish as a mother tongue learn to pay special attention to Actor entities, leaving the establishment of cohesion to communicative inferences. For an adult learner of a German with Spanish as a first language it is thus necessary to recognise the patterns of conceptualisation which correspond to the target language (that is, the salience of a global topic entity) and at the same time, to give up the established native pattern of conceptualisation (which implies regarding agentivity as primary for salience). Therefore, the acquisition of the conceptual patterns of the target language is connected with the acquisition of the formal requirements for the production of a cohesive text in the target language.

. L1 Spanish – L2 German As mentioned above, the learner group was divided into three levels, according to their performance of the target language. As a consequence, and due to the small number of relevant utterances, the results obtained where not sufficient for a statistical analysis. Therefore we will concentrate on a qualitative analysis. The first group consists of three less advanced speakers (out of the total 15). These speakers basically stick to the reference tracking system of the source language: The inanimate Actor entity is coded as subject in a pre-verbal topical position. This can be seen in examples (8) to (10): (8) die Blätter haben ihn gestoßen the sheets have him pushed ‘the sheets have pushed him’

JB[v.20020404] Prn:26/06/2003; 13:36

F: SIB2611.tex / p.14 (720-769)

 Jorge Murcia-Serra

(9) einige Steine haben ihn geschlagen some stones have him hit ‘some stones have hit him’ (10) die Maschinen die zerdrücken ihn the machines they squash him ‘the machines squash him’ As can be see, this group of speakers has not recognised the cohesive function of the subject as a result of global text planning, since they systematically set inanimate Actor entities as the topic at a local utterance level. The second group of learners contains six speakers. This group shows a somehow mixed picture in that they set an inanimate Actor in the subject function as a local topic in more or less half of the cases in which the global topic entity carries the role of an Undergoer, while in the rest of the cases they use passive constructions. This can be seen in examples (11) and (12) which stem from to the same speaker: (11) 1

2

3

(12) 1

2

3

4

5

und er fängt dann in den Sand zu graben and he starts then in the sand to dig ‘and then he starts digging in the sand’ und auf einmal sieht so aus, als ob der Sand sich öffnet and at once looks like as if the sand itself opens ‘and suddenly it looks as if the sand would get appart’ und er wird geschluckt and he gets swallowed ‘and he is swallowed down’ und er fängt an schneller zu arbeiten and he starts faster to work ‘and he starts to work faster’ um das Loch größer zu machen so-as-to the hole bigger to make ‘in order to make the hole bigger’ aber er hat keine Chance praktisch but he has no chance actually ‘but he has actually no chance’ und die Wände kommen zu ihm and the walls come to him ‘and the walls get closer to him’ und zerknutschen ihn (wrong word used: intended was zerquetschen = squash)

JB[v.20020404] Prn:26/06/2003; 13:36

F: SIB2611.tex / p.15 (769-819)

Acquiring the linkage between syntactic, semantic and informational roles 

and neck him ‘and squash him’ Clearly, this group has begun to recognise the function of the subject category for global topic reference but still is in an unstable stadium of acquisition. This is clearly shown by the following example of a self-repair: (13) 1

2

und diese zwei Maschinen machen ihn dann ganz zu Sand and these two machines turn him then totally to sand ‘and then these two machines turn him completely into sand’ also sein Körper wird dann zu Sand gemacht that-is his body gets then to sand turned ‘that is, his body gets then turned to sand’

In utterance no. 1 the inanimate Actor entity is formulated as the topic of the utterance. This is corrected in utterance no. 2 thanks to the target-like usage of a passive construction. The last group of seven speakers show in all relevant cases a target consistent linking subject–global topic as showed in example (14): (14) 1

2

3

4

5

6

7

dann entdeckt er ein Loch then discovers he a hole ‘then he discovers a hole’ wo er viel Wasser sehen kann where he lots-of water see can ‘where he can see a lot of water’ ich glaube das ist das Meer I think that is the sea ‘I think, it is the sea’ dann rennt er dahin then runs he there-to ‘and he runs to that place’ und versucht ein Loch zu machen oder zu graben and tries a hole to make or to dig ‘and tries to make or to dig a hole’ um ans Wasser zu gelangen in-order-to to-the water to reach ‘in order to reach the water’ aber er schafft es nicht ganz but he doesn’t quite manage ‘but he doesn’t quite manage’

JB[v.20020404] Prn:26/06/2003; 13:36

F: SIB2611.tex / p.16 (819-873)

 Jorge Murcia-Serra

8

und wird von einer riesigen Maschine zerquetscht and gets by a huge machine squashed ‘and gets squashed by a huge machine’

. Conclusions The results obtained show that the use of the category subject as a coding for the global topic entity required in the target language seems to pose important problems for learners of German with Spanish as a mother tongue. The acquisition of such a linking not only requires the recognition of the function of the category subject for textual cohesion but also of the patterns of conceptualisation required for speaking in the target language. This requires in the case of narratives the focusing of attention on the protagonist of the events being presented. At the same time, it is necessary to diminish the attention given to Actor entities in the source language. As we have seen, even advanced learners of German keep to the conceptualisation patterns of the Spanish language. The structures observed for the first and partly for the second group of learners cannot be attributed to the lack of the acquisition of the passive as such. On the one side, as it was mentioned, passive constructions are also existing in Spanish. On the other side it must be taken into account, that all speakers studied here had learned the German language in a classroom context and consequently had had plenty of drilling in the use of passive in German. This means, that the problem is not the acquisition of the formal means, but basically one of the appropriate usage of such means paired with the corresponding patterns of conceptualisation. Nevertheless, the group of most advanced or native-like speakers could be shown to have acquired the relevant linking of syntactic, semantic, and informational roles and relations and consequently the acquisition of both the target information organisation and the conceptualisation of events for verbalisation. It is not surprising to see that this is accompanied by the mastering of the target rules in relation to the fixed position of the verb in the sentence (V2) and the connected subjacency of the subject in German. At the same time, the group of native-like speakers does not show utterances with null-subjects (in contrast to the speakers at the lower levels). As we saw, a possible function of the linking in Spanish of subject and Actor is to facilitate the identification of the referent in cases of subject drop (which are the rule). However, the attained acquisition is surely not the consequence of class-room teaching, since the patterns of information organisation and focus of attention described here are in general poorly understood and consequently not taught in class. The production of utterances in discourse is a complex task involving not only the

JB[v.20020404] Prn:26/06/2003; 13:36

F: SIB2611.tex / p.17 (873-921)

Acquiring the linkage between syntactic, semantic and informational roles 

uttering of grammatical sentences in the concerning language, but also the organisation of information in the text in a determined way according to the systematic patterns of conceptualisation in that language. These patterns in turn are related, as we saw, to determined structural regularities in the language (cf. Carroll 1993, 1997, 1999; Carroll & Stutterheim 1993, 2003; Stutterheim 1999). The trigger for the successful acquisition of the patterns referred to here seems to be frequent usage by the learners of the target language in a communicative context where longer texts are heard and produced. This offers the necessary input and the necessary practice for the learner in order to acquire the relevant patterns of information organisation and the connected focus of attention. It seems thus to be the case, that the relevant patterns of conceptualisation and grammatisation are not necessarily an insurmountable hurdle for the successful second language acquisition by adults, so that “thinking for speaking” in the foreign language can be attained and could probably also be taught (cf. J. B. Carroll 1979).7 The condition for this is that the relevant patterns of conceptualisation especially in relation to the production of texts (and not only of single utterances) be identified.

Notes * This research was founded by the German Research Society (DFG). I would like to thank Mary Carroll, Alicia González de Sarralde, Christine Dimroth, and Marianne Starren for their comments on an earlier version of this paper. . In the following examples, bold letters are used to indicate the topic of the utterance, while underlining indicates the agent entity. . Other coding possibilities for topic-information, like intonation or special particles will not be considered in this paper. . This could also be disputable (cf. Reis 1982; Contreras 1976: 121ff.). But this is not the place for such a discussion, and as it will be seen, for the purpose of the present study, the definition given here is fully defendable. . Following Van Valin and LaPolla (1997), in the following we will write the macroroles Actor and Undergoer in capitals to distinguish them from “plane” thematic roles, like agent or patient. . The actual affiliation to one or the other macrorole might be language dependent (cf. Van Valin & LaPolla 1997: 143). . Distributed by Katholisches Filmwerk GmbH, Ludwigstr. 33, 60327 Frankfurt am Main, Germany. . Nevertheless, there are some conceptual categorisations in language, like spatial and temporal relationships, which seem to pose special problems for adult second language acquisi-

JB[v.20020404] Prn:26/06/2003; 13:36

F: SIB2611.tex / p.18 (921-1014)

 Jorge Murcia-Serra

tion even in the case of native-like speakers (cf. Carroll & Lambert, this volume; Carroll et al. 2000).

References Bock, J. K. & Warren, R. K. (1985). Conceptual accessibility and syntactic structure in sentence formulation. Cognition, 21, 47–67. Bock, K. & Levelt, W. (1994). Language Production: Grammatical encoding. In M. A. Gernsbacher (Ed.), Handbook of Psycholinguistics (pp. 945–984). San Diego: Academic Press. Bock, K., Loebell, H., & Morey, R. (1992). From conceptual roles to structural relations: bridging the syntactic cleft. Psychological Review, 99(1), 150–171. Carroll, J. B. (1979). Linguistic relativity, contrastive linguistics and language learning. In D. Nehls (Ed.), Studies in Contrastive Linguistics and Error Analysis. I. The theoretical background (pp. 7–30). Heidelberg: Julius Groos Verlag. Carroll, J. M. (1958). Process and content in psycholinguistics. In R. Glaser (Ed.), Current Trend in the Description and Analysis of Behabior (pp. 175–200). Pittsburgh: University of Pittsburgh Press. Carroll, M. (1993). Deictic and intrinsic orientation in spatial descriptions: A comparison between English and German. In J. Altarriba (Ed.), Cognition and Culture: A CrossCultural Approach to Cognitive Psychology (pp. 23–44). Amsterdam: Elsevier. Carroll, M. (1997). Changing place in English and German: language-specific preferences in the conceptualization of spatial relations. In J. Nuyts & E. Pederson (Eds.), Language and Conceptualization (pp. 137–161). Cambridge: Cambridge University Press. Carroll, M. (1999). Representing path in language production in English and German: Alternative perspectives on Figure and Ground. In C. Habel & C. v. Stutterheim (Ed.), Räumliche Konzepte und sprachliche Strukturen. Tübingen: Niemeyer. Carroll, M. & Lambert, M. (this volume). Information structure in narratives and the role of grammaticised knowledge: A study of adult French and German learners of English. Carroll, M., Murcia-Serra, J., Watorek M., & Bendiscioli A. (2000). The relevance of information organisation to second language acquisition studies: the descriptive discourse of advanced adult learners of German. In Studies in Second Language Acquisition, (Special issue edited by C. Perdue) 22(3), 441–466. Carroll, M. & Stutterheim, C. von (1993). The representation of spatial configurations in English and German and the grammatical structure of locative and anaphoric expressions. Linguistics, 31(6), 1011–1041. Carroll, M. & Stutterheim, C. von (2003). Typology and information organisation: the question of perspective taking and and language-specific effects in the construction of events. In A. Ramat (Ed.), Typology and Second Language Acquisition. Berlin: de Gruyter. Contreras, H. (1976). A theory of Word Order with Special Reference to Spanish. Amsterdam: North Holland. Dowty, D. (1991). Thematic Proto-roles and argument selection. Language, 67(3), 547–619.

JB[v.20020404] Prn:26/06/2003; 13:36

F: SIB2611.tex / p.19 (1014-1105)

Acquiring the linkage between syntactic, semantic and informational roles 

Fernández Lagunilla, M. & Anula Rebollo, A. (1995). La oración simple. In M. Fernández Lagunilla & A. Anula Rebollo (Eds.), Sintaxis y cognición. Introducción al conocimiento, el procesamiento y los déficits sintácticos (pp. 255–308). Madrid: Síntesis. Fillmore, C. J. (1977). The case for case reopened. In P. Cole & J. M. Sadock (Eds.), Syntax and Semantics. Vol. 8. Gramatical Relations (pp. 59–81). New York: Academic Press. Flores d’Arcais, G. B. (1975). Some perceptual determinants of sentence construction. In G. B. Flores d’Arcais (Ed.), Studies in Perception: Festschrift for Fabio Metelli (pp. 344–373). Milan: Martello-Giunti. Givón, T. (1983). Topic continuity in discourse: An introduction. In T. Givón (Ed.), Topic continuity in Discourse: A Quantitative Cross-language Study (pp. 5–41). Amsterdam: John Benjamins. Givón, T. (1992). The grammar of referential coherence as mental processing instructions. Linguistics, 30, 5–55. Givón, T. (1994). The pragmatics of de-transitive voice: Functional and typologiyal aspects of inversion. In T. Givón (Ed.), Voice and Inversion (pp. 3–44). Amsterdam: John Benjamins. González de Sarralde González, A. (2000). Subjektpositionierung in Erzählungen. Versuch einer kognitiv-funktionalen Erklärung anhand mündlicher narrativer Texte des Spanischen. Frankfurt am Main: Peter Lang. Heim, I. (1982). The Semantics of Definite and Indefinite Noun Phrases. University of Massachusetts. Ph. D. Dissertation. Igoa, J. M. (1996). The relationship between conceptualization and formulation processes in sentence production: Some evidence from Spanish. In M. Carreiras, J. E. García-Albea, & N. Sebastián-Gallés (Eds.), Language Processing in Spanish (pp. 305–351). Mahwah, NJ: Lawrence Erlbaum Associates. Jaeggli, O. (1986). Arbitrary plural pronominals. Natural Language and Linguistic Theory, 4, 43–76. Keenan, E. L. (1976). Towards a universal definition of ‘subject’. In C. Li (Ed.), Subject and Topic (pp. 303–333). New York: Academic Press. Kuno, S. & Kaburaki, E. (1977). Empathy and syntax. Linguistic Inquiry, 8, 627–672. Kuno, S. (1983). Functional Syntax: Anaphora, Discourse and Empathy. Chicago: University of Chicago Press. Lambrecht, K. (1994). Information Structure and Sentence Form. Cambridge: University Press. Levelt, W. J. M. (1989). Speaking: From Intention to Articulation. Cambridge, MA: MIT Press. Murcia-Serra, J. (2001). Grammatische Relationen im Deutschen und Spanischen: Eine empirische Untersuchung zur Rolle der einzelsprachlichen Form bei der Konzeptualisierung von Äußerungen im Text. Frankfurt am Main: Peter Lang. Neumann-Holzschuh, I. (1997). Die Satzgliedanordnung im Spanischen: Eine Diachronische Analyse. Tübingen: Niemeyer. Osgood, C. E. & Bock, J. K. (1977). Salience and sentencing: some production principles. In S. Rosenberg (Ed.), Sentence production: Developments in research and theory (pp. 89–140). Hillsdale, NJ: Lawrence Erlbaum Associates.

JB[v.20020404] Prn:26/06/2003; 13:36

F: SIB2611.tex / p.20 (1105-1205)

 Jorge Murcia-Serra

Primus, B. (1993). Word order and information structure: A performance-based account of topic positions and focus positions. In J. Jacobs, A. von Stechow, W. Sternefeld, & T. Venneman (Eds.), Syntax : Ein internationales Handbuch zeitgenössischer Forschung, Vol. 1 (pp. 880–896). Berlin: de Gruyter. Reinhart, T. (1982). Pragmatics and linguistics: an analysis of sentence topics. Bloomington, IN: IULC. (First published 1981 in Philosophica, 27, 53–94.) Reis, M. (1982). Zum Subjektbegriff im Deutschen. In W. Abraham (Ed.), Satzglieder im Deutschen (pp. 171–211). Tübingen: Narr. Sasse, H. J. (1987). The thetic/categorial distinction revisited. Linguistics, 25, 511–580. Slobin, D. (1991). Learning to think for speaking: native language, cognition, and rhetorical style. Pragmatics, 1, 7–25. Slobin, D. (1996). From ‘thought and language’ to ‘thinking for speaking’. In J. J. Gumperz & S. C. Levinson (Eds.), Rethinking linguistic relativity (pp. 70–96). Cambridge: Cambridge University Press. Sridhar, S. N. (1988). Cognition and Sentece Production: A Cross-Linguistic Study. Heidelberg: Springer-Verlag. Stutterheim, C. von (1997). Einige Prinzipien des Textaufbaus Empirische Untersuchungen zur Produktion mündlicher Texte. Tübingen: Niemeyer. Stutterheim, C. von (1999). How language specific are processes in the conceptualiser? In R. Klabunde & C. von Stutterheim (Eds.), Representation and Processes in Language Production (pp. 153–179). Wiesbaden: Deutscher UniversitätsVerlag. Suñer, M. (1983). Proarb . Linguistic Inquiry, 14, 188–191. Talmy, L. (1978). Figure and ground in complex sentences. In J. Greenberg, C. Ferguson, & E. Moravcsik (Eds.), Universals of Human Language, Vol. 4 (pp. 625–649). Stanford: Stanford University Press. Talmy, L. (1988). The relation of grammar to cognition. In B. Rudzka-Ostyn (Ed.), Topics in Cognitive Linguistics (pp. 165–205). Amsterdam: John Benjamins. Tannenbaum, P. H. & Williams, F. (1968). Generation of active and passive sentences as a function of subject and object focus. Journal of Verbal Learning and Verbal Behavior, 7, 246–250. Tomlin, R. S. (1995). Focal attention, voice, and word order: an experimental, crosslinguistic study. In P. Downing & M. Noonan (Eds.), Word Order in Discourse (pp. 517–554). Amsterdam: John Benjamins. Tomlin, R. S. (1997). Mapping conceptual representations into linguistic representations: the role of attention in grammar. In J. Nuyts & E. Pederson (Eds.), Language and Conceptualisation. Cambridge: Cambridge University Press. Turner, E. A. & Rommetveit, R. (1968). Focus of attention in recall of active and passive sentences. Journal of Verbal Learning and Verbal Behavior, 7, 543–548. Ulrich, M. (1985). Thetisch und Kategorisch – Funktionieren der Anordnung von Satzkonstituenten am Beispiel des Rumänischen und anderer Sprachen. Tübingen: Narr. Vallduví, E. (1992). The Informational Component. New York: Garland. Vallduví, E. (1993). Information packaging: A survey. Research Paper HCRC / RP-44. Edinburg: HCRC Publications. Van Valin, R. D. Jr. & LaPolla, R. J. (1997). Syntax. Structure, meaning and function. Cambridge: Cambridge University Press.

JB[v.20020404] Prn:26/06/2003; 13:36

F: SIB2611.tex / p.21 (1205-1209)

Acquiring the linkage between syntactic, semantic and informational roles 

Whorf, B. J. (1956). Language, Thought, and Reality. Selected writings edited and with an introduction by J. B. Carroll. Cambridge, MA: The Technology Press of MIT.

JB[v.20020404] Prn:26/06/2003; 13:43

F: SIB2612.tex / p.1 (35-98)

Gestures, referents, and anaphoric linkage in learner varieties Marianne Gullberg

.

Introduction

For discourse to be comprehensible, referents have to be uniquely identifiable. This paper is concerned with how gestures contribute to this process in interaction with speech in learner discourse. It is a common lay assumption that gestures are chiefly connected to exophoric, demonstrative expressions, such that gestures refer directly to and index entities in the real world. A typical example would be a speaker pointing to an object saying, “look at that”. However, speakers’ gestures also depict and index such abstract things as discourse referents and discourse. Gestures can be endophoric, and, more specifically, anaphoric. The systematic association between gestures, space, and speech allows cohesion to be established in all anaphoric domains. The purpose of this paper is to outline how such gestural cohesion can shed light on the specific characteristics of anaphoric linkage in learner varieties. Target language-specific preferences for anaphora, cohesive patterns and information organisation in speech are acquired late, if at all, by adult language learners. Difficulties can be found in all anaphoric domains: space, time, and referential movement of first order entities. In the domain of referential movement, i.e. the movement of animate entities with a high degree of control over events and actions in discourse (Klein & Perdue 1997; von Stutterheim & Klein 1989) studies show that learner varieties at early non-finite stages of proficiency (the Basic Variety stage, Klein & Perdue 1997) typically rely heavily on implicit information established in the surrounding context. In contrast, at later stages learner varieties are more explicit than native varieties. In particular, the domain of maintained reference, in which agents typically move from topic to topic in successive utterances, is characterised by full NPs being favoured over pronouns and zero-anaphora even in contexts where referents are supposedly both accessible and recoverable (e.g. Ahrenholz 1998; Carroll et al. 2000; Extra, Strömqvist, & Broeder 1988; Givón 1984; Hendriks 2000; Muñoz 2000; Strömqvist & Day 1993).

JB[v.20020404] Prn:26/06/2003; 13:43

F: SIB2612.tex / p.2 (98-159)

 Marianne Gullberg

(1) elle a eh + donné + pour eh la dame ‘she has uh + given + to uh the lady et ehm + la dame allé pour une autre dame + eh + and uhm + the lady went to another lady’ In (1), a learner of French2 introduces the referent la dame in object position in the first utterance, and then uses the same NP in subject position in the immediately subsequent utterance instead of demoting the referent using a pronoun. While there may be pragmatic reasons in L1 for using a full NP to refer to an immediately preceding referent, learners seem to routinely use NPs in such contexts regardless of appropriateness. The reasons for this over-extension of full nominal forms have been amply discussed. It has been suggested that, since lexical means are acquired before grammatical ones in all areas of acquisition, so lexical means of co-reference should be expected before more grammatical, pronominal ones. The pronominal forms of anaphora are also particularly complex as they encode multiple grammatical categories simultaneously, typically gender, number, and case. This complexity makes pronominal systems error-prone (Hendriks 2000). It has therefore been suggested that learners shun them in favour of full nominal expressions as a way of avoiding ambiguous reference created by erroneous pronominal forms (Williams 1988). Moreover, the choice of form depends on the intricate interplay between information organisation at a local level (e.g. given vs. old information) and at a global level beyond the single utterance (Carroll & Lambert, this volume; Carroll et al. 2000). This span over multiple levels clearly represents a challenge to learners whose processing capacities are already occupied by formal aspects at a local level, and may further motivate avoidance of pronouns. The influence of the first language (Jin 1994), of markedness factors (Muñoz 1995), and task (Muñoz 2000) have also been explored. In a more interactional perspective, it has been proposed that the use of full NPs is motivated by the social co-ordination of talk in progress. Full NPs can be used as a means to mark disagreement or manage viewpoint (Fox 1987; Pekarek 1999). This paper will outline how gestural cohesion can shed light on the effects and causes of the specific characteristics of anaphoric linkage in spoken learner varieties. First, it will be shown that spoken learner varieties come with particular gestural profiles that are related to the characteristics of spoken varieties in non-trivial ways. Put differently, there is a learner variety of gestural anaphoric linkage which (a) differs systematically from the L1 variety in the same way the spoken learner variety differs from the native variety, and which (b) is subject to developmental change along the same lines. Second, the effects of and two possible explanations for this learner-specific variety of gestural anaphoric linkage will be discussed.

JB[v.20020404] Prn:26/06/2003; 13:43

F: SIB2612.tex / p.3 (159-203)

Gestures, referents, and anaphoric linkage in learner varieties

. An introduction to gestures and gestural cohesion Gestures are defined in this paper as co-speech gestures or the (manual) movements speakers perform unwittingly while they speak as part of the expressive effort (cf. Kendon 1993; McNeill 1992). These movements are closely and systematically related to language and speech. Gestures are semantically co-expressive with speech, such that they often convey meaning also present in speech either iconically, or by way of spatial contiguity or indexicality. In (2), a speaker is performing a two-handed gesture which outlines a rectangular shape in space, iconically depicting the shape of the reception being talked about (see also Figure 2.2 in Section 5). (2) la femme qui est [+ dans la réception +] ‘the woman who is in the reception’ In addition to the overlap in content, the onset and duration of a gesture is tightly co-ordinated with the onset and duration of the referential expression in speech to which it is semantically linked. The square brackets in (2) mark the total extension of the manual movement, from preparation to retraction; boldface marks the stroke or the most forceful movement of the gesture; the underlined part marks a hold, i.e. a phase where the hands are held still in the air with maintained handshape (McNeill 1992); the plus (+) indicates a (non-measured) pause. As can be seen in (2), the hands move into position before la réception is uttered. The actual gestural movement occurs as the speaker is saying dans la, but the hands are maintained in the air until the noun réception has been uttered. Only then do the hands come down to rest. Without going into technical details, this tight semantic and temporal coordination suggests interdependence between the modalities. Gestures and space also offer specific possibilities for cohesion. The mechanism of gestural cohesion rests on the continued or recurring gestural patterns – handedness, hand configuration, or specific spatial area – that manifest themselves over a stretch of discourse (labelled “catchments” by McNeill 2000) to reflect consistent visuospatial imagery running through a discourse segment. The close semantic and temporal co-ordination of gestures and speech also allow gestural patterns to be repeatedly associated with a given referential expression in speech (Levy & McNeill 1992; McNeill & Levy 1993). The association can be accomplished through iconicity or through indexicality such that a hand or a particular hand shape can come to represent a specific character throughout a story (Kendon 1972; McNeill & Levy 1993). Space may also be associated with particular referential expressions, such that one side of space is consistently indicated when speaking of France and the other when speaking of Germany, for instance. These different features are typically conflated such that spatial area and handedness coincide. Once



JB[v.20020404] Prn:26/06/2003; 13:43

F: SIB2612.tex / p.4 (203-258)

 Marianne Gullberg

the association between a referent and a gesture-space complex is established, the activation of the spatial area or the hand (shape) results in the activation of the associated referent in discourse.

. Data background The analyses in this paper are based on cross-sectional data of video recorded story retellings from five native speakers of Swedish learning French as a foreign language in a classroom setting, and five native speakers of French learning Swedish as a foreign language under similar circumstances. The subjects retold a cartoon story in both their native and their foreign language to a native speaker of the respective languages. Half of the subjects performed the task in L1-L2 order, the other half in L2-L1 order. The subjects thus acted as their own controls with respect to gesture production as well as with regard to story telling skills. The subjects were all at an intermediate level as established by a panel of native speakers, or at an early postbasic variety level (cf. Klein & Perdue 1997). This latter characterisation is based on the observation of some verbal and nominal morphology in the learner speech (typically finite forms of lexical verbs in the present tense, and some base form for past tense, tentative definiteness, gender, and number markings). This grouping notwithstanding, there was individual variation in the actual proficiency level. In addition to this data set, two learners from the Dutch and Swedish parts of the European Science Foundation project on adult second language acquisition have been analysed for gesture production.1 These data serve as an important source of information and confirmation regarding a different learner type (untutored), a different proficiency level (lower), a different discourse type (personal narrative), and different source- and target languages (L1 Finnish > L2 Swedish, and L1 Moroccan Arabic > L2 Dutch). Note, however, that since there are no native baseline data for the gesture performance for the ESF data, they have not been included in the brief quantitative outline in Sections 4 and 5. In the native and non-native data sets animate and inanimate referents were coded for referential status at the utterance level as introduced (new), maintained (old), re-introduced (new). Introduced refers to the first mention of an argument independently of its position in the clause (subject, object, or oblique argument). Maintained refers to the maintenance of the same referent in subject position in a subsequent clause. Re-introduced refers to the re-introduction of a previously mentioned referent subsequent to a clause containing a different subject. Referents were also coded for whether they contributed to topic or focus elements. Topics were identified on the basis of the aboutness criterion: “The topic of a sentence is the thing which the proposition expressed by the sentence is ABOUT”

JB[v.20020404] Prn:26/06/2003; 13:43

F: SIB2612.tex / p.5 (258-343)

Gestures, referents, and anaphoric linkage in learner varieties

(Lambrecht 1994). Focus was defined in Lambrecht’s terms as “that portion of a proposition which cannot be taken for granted at the time of speech. It is the UNPREDICTABLE or pragmatically NON-RECOVERABLE element in an utterance. The focus is what makes an utterance into an assertion” (1994: 207). Note that topic-comment structure is not dependent on a referent’s status as givennew, but rather reflects the underlying question that a text sets out to answer (von Stutterheim & Klein 1989). All gestures in the native and non-native data were identified. Those gestures that co-occurred with referential expressions in speech were coded for location in gesture space. Note that gestures occurring with the predication of an utterance or the verb were not considered. Note also that in the L2 data, the same referential expression is often repeated with a simultaneous repetition of the gesture. Such speech-gesture clusters were only considered once.

. Gestural anaphoric linkage in L1 The general mechanism of gestural cohesion was briefly outlined above. The specifics of anaphoric linkage can be observed in fine-grained detail in the domain of referential movement in native production. When a referent is first introduced in discourse, speakers can accompany the NP with a pointing, deictic gesture, or with an iconic gesture depicting the referent. The gesture associates the referent with or localises it in the part of space indicated by the point or in the place where the iconic gesture was performed.2 In Figure 1, a native speaker of Swedish localises both the girl at the counter and the manager at their first mention. The girl is localised deictically, whereas the manager is anchored iconically such that a rectangular shape is outlined to symbolise his location. Speakers may point back to the locus in space previously established as associated with the referent when the referent is next mentioned. The anaphoric gesture re-activates the location and the referent associated with it (Levy & McNeill 1992; Marslen-Wilson, Levy, & Komisarjevsky Tyler 1982; McNeill & Levy 1993). The association between a referent and a locus serves not only to establish the referent, but rather to establish the larger scenes or spatial settings that the referent occupies. Indeed, when a localised referent moves, speakers often track it gesturally across gesture space, such that the end point of the trajectory becomes the referent’s new anchor point. Notice, however, that such tracking gestures typically occur with the predication or the verbal element of an utterance and not with referring expressions (Levy & McNeill 1992). As these anchor points multiply, speakers charge concrete gesture space with referential meaning and turn it into a map of discourse (Liddell 1996; McNeill & Pedelty 1995). The relative location of a referent is usually irrelevant,



JB[v.20020404] Prn:26/06/2003; 13:43

F: SIB2612.tex / p.6 (343-383)

 Marianne Gullberg

1 så visar hon de eh till eh då en [tjej] i disken 1 ‘so she shows it uh to uh a [girl] at the counter

2 + som tar detta då å tycker att de ser mycke mystisk ut + varefter hon tar de till [typ föreståndarn] 2 + who takes this then and thinks it looks very strange + whereupon she takes it to [sort of manager]’

* Only illustrated gestures are marked in the transcription

Figure 1. A native speaker of Swedish localises two referents at their introduction.* [] indicates the gesture phrase, boldface marks the gesture stroke, underlined marks gesture hold, and + indicates non-measured pause. Table 1. Number of gestures occurring with introduced, maintained or re-introduced referential expressions in L1.

L1

Introduced

Maintained

Re-introduced

43/171 (25%)

3/285 (1%)

18/128 (14%)

but once it has been established, it is constant and referential space takes on an absolute topological quality. This referential space or discourse map allows explicit and unequivocal visual co-reference. A number of studies have shown that the gestural anchoring of referents is determined by their referential status as new or given, and on their status in the information structure. New (or re-introduced) referents which are part of focus elements tend to be localised, whilst given (maintained) referents which are part of topic elements tend not to be (Levy & McNeill 1992; Marslen-Wilson et al. 1982). These findings were replicated in this study (see Table 1). The speaker in Figure 1 gesturally anchors only the new referents; he does not refer back to them while they are maintained. A similar effect for referential status has been observed for Sign Language, where concrete, specific referents with high thematic value are more likely to be localised than abstract or concrete referents with low thematic value (Engberg-

JB[v.20020404] Prn:26/06/2003; 13:43

F: SIB2612.tex / p.7 (383-499)

Gestures, referents, and anaphoric linkage in learner varieties

Pedersen 1993). This relationship between gestural cohesion and discourse status is further seen in the developmental effects of the deployment of anaphoric gestures. Only when children master the principles of discourse construction, information structure, and the use of appropriate grammatical means to organise these do they start producing gestures to mark referents in the adult fashion. In fact, they do not appear to produce adult-like gestural anaphora before the age of 12 (McNeill 1986).

. Gestural anaphoric linkage in L2 The results from studies of native discourse indicate that the deployment of anchoring and/or anaphoric gestures in L1 crucially depends on the very factors that cause learners problems in L2, viz. referential status and information organisation beyond the single utterance level. What does this mean for learners’ use of anaphoric gestures in L2 production? In Figure 2 a learner retells exactly the same episode in the story as in Figure 1, but this time in his foreign language. The two new referents in this episode are treated as in L1: they are introduced with full NPs and gesturally anchored (la femme in Figure 2.1, and le supervisé in 2.6). In contrast to the native example, both these NPs are further specified by locative expressions which are also gesturally expressed and spatially anchored; (la femme) dans la réception in 2.2, and (le supervisé) de la pharmacie in 2.7. The referent la femme is subsequently maintained throughout this stretch by use of full NPs. With every such full NP, the learner points back to the anchor point associated with the referent. Notice that there are two female characters in this section. Since the first one is here labelled with a pronoun, elle in 2.1, the NP in Figure 2.3 could have been taken to re-introduce this first character. However, the anaphoric gestures occurring with the NPs consistently indicate the very location in space where the first NP was anchored, thereby indicating the co-referential reading. This example thus illustrates three points. First, learners over-mark referents gesturally in the sense that referents are first located at their introduction and then anaphorically indicated at every subsequent mention, even in an immediate context. Moreover, even at their introduction referents can be gesturally “over-marked”. The specifying locative expressions (dans la réception) are sometimes also spatially anchored such that a single referent is gesturally marked twice. Second, referents that are overmarked in gesture are also over-marked in speech by nominal expressions. Maintained referents, for instance, are typically expressed with full NPs and also with anaphoric gestures. There is thus simultaneous over-marking in both modalities. Third, gestural anaphoric linkage allows unequivocal visual co-reference. Tables



JB[v.20020404] Prn:26/06/2003; 13:43

F: SIB2612.tex / p.8 (499-566)

 Marianne Gullberg

1 eh elle a un une script + qui elle eh+ donné à la [+ femme+] 1 ‘uh she has a a prescription + that she uh + give to the [+ woman+]

2 qui est [dans la réception+]

3 et [la femme] ne comprend pas +

4 et [la femme] +

2 who is [in the reception +]

3 and [the woman] doesn’t understand +

4 and [the woman] +

5 [dans le réception]

6 eh donné la script eh de [+ le + supervisé] 6 uh give the prescription uh of [+ the + supervisor]

7 + [de la pharmacie]

5 [in the reception]

7 + [of the pharmacy]’

Figure 2. A learner of French localises referents gesturally in space at their introduction and at all subsequent mentions.

2–4 illustrate the first two points by summarising some differences between the patterns of gestural anaphoric linkage in L1 and L2 in the Swedish-French data. While native speakers gesturally localise both newly introduced referents, and re-introduced referents, learners perform gestural anaphora with new, re-

JB[v.20020404] Prn:26/06/2003; 13:43

F: SIB2612.tex / p.9 (566-567)

Gestures, referents, and anaphoric linkage in learner varieties 

Table 2. Number of gestures occurring with introduced, maintained or re-introduced referential expressions in L1 vs. L2.

L1 L2 L1 vs. L2

Introduced

Maintained

Re-introduced

43/171 (25%) 58/149 (39%) χ2 = 3.2, p = 0.0736

3/285 (1%) 31/208 (15%) χ2 = 28.85, p≤ 0.0001

18/128 (14%) 26/136 (26%) χ2 = 0.59, p = 0.4424

Table 3. Number of gestures occurring with topic or focus parts in L1 vs. L2.

L1 L2 L1 vs. L2

Topic

Focus

10/64 (16%) 41/115 (36%) χ2 = 4.03, p = 0.0447

54/64 (84%) 74/115 (64%) χ2 = 1.05, p = 0.3055

Table 4. Number of gestures occurring with NPs or pronouns in L1 vs. L2.

L1 L2

NP

Pron

59/64 (92%) 108/115 (94%)

5/64 (8%) 7/115 (6%)

introduced, and maintained referents. When the distribution of anaphoric gestures is considered across topic vs. focal parts of utterances, a similar pattern is seen. While the majority of the gestures occurring with referential expressions are part of the focus both in native and non-native discourse, learners perform a significantly greater number of gestures on topic material than native speakers. Finally, gestures occurring with referential expressions largely coincide with NPs both in native and non-native discourse. This means that in (nearly) all cases where anaphoric gestures occur, the indicated referents are expressed as NPs in speech. For native speech, this means with new or re-introduced referents in focus. For learner varieties, however, it means that in addition to contexts of new and re-introduced referents in focus, gestures occur with NPs expressing maintained referents in topic (Levy & McNeill 1992). In other words, just as learners’ speech is characterised by nominal over-explicitness for maintained referents, so their gestural performance is over-marked with respect to the same expressions. Note that the same features can be observed in the two untutored learners in the ESF data set although they have not been included in the quantitative analyses. Despite the different discourse type (personal narrative), proficiency level (lower), and languages involved the tendency to gesturally refer to nominally maintained referents is the same.

JB[v.20020404] Prn:26/06/2003; 13:43

F: SIB2612.tex / p.10 (567-668)

 Marianne Gullberg

An additional observation is that there is a clear tendency in the data for a reduction of gestural anaphora with the presence of pronominal means (including zero anaphora) of co-reference. The decrease in localising gestures is not a matter of fluency or of lexical development. The learner in Figure 2 is relatively fluent in the sense that his hesitations are few and brief, and his use of the limited lexicon is quite creative. But the absence of pronouns is conspicuous, and his persistent use of NPs for maintained referents is still accompanied by localising gestures. Inversely, learners with relatively greater fluency problems but who do use anaphoric pronouns for maintained referents do not display anaphoric gestures with these expressions. The amount of localising-anaphoric gestures appears to depend crucially on the development of those grammatical means that allow a learner to mark referents as already established. As a side remark, the tight link between gestures and NPs may at first sight seem surprising. In recognition of the powerful effect of visual co-reference, it is often assumed that gestures should function to disambiguate vague or underspecified expressions such as pronouns or zero anaphora.3 In view of learners’ poor command of pronouns and zero anaphora, one might therefore have assumed that where (erroneous) pronouns occur, they would be in need of gestural support for anaphoric resolution. However, gestures do not in general co-occur with pronouns. The only time pronouns are accompanied by anaphoric gestures in native or learner language is when the pronouns receive emphatic stress, or when the pronominal form itself is being negotiated, as in (3). (3) [hans do] doktor eh [hennes] eh non hans hennes eh hans [doktor] ‘[his do] doctor uh [her] uh no his her uh his [doctor]’ In this case, the gesture does not serve to identify a discourse referent. Instead, the gesture serves as a placeholder for or a localisation of form. The negotiation of the form is conducted at a meta-discursive level. The gesture is therefore not anaphoric at the narrative level, but does in fact serve to localise form at an even more abstract level of space. To summarise thus far, the characteristic pattern for anaphoric linkage in the spoken learner variety is mirrored in gesture: maintained referents in topic are over-marked in speech by full NPs and in gesture by consistent anaphoric pointing. Put differently, the learner variety of gestural cohesion, or the gestural correlate to the spoken learner variety, differs systematically from L1 production in precisely the same way as the spoken variety. It is over-explicit, specifically with regard to maintained referents in topics. Moreover, it changes with the development of the grammatical means to achieve maintained reference. What, then, are the effects of the learner variety of gestural anaphoric linkage and what motivates it?

JB[v.20020404] Prn:26/06/2003; 13:43

F: SIB2612.tex / p.11 (668-684)

Gestures, referents, and anaphoric linkage in learner varieties

. Gestural anaphoric linkage – Is it communicatively motivated? It has been argued that learners favour full NPs to avoid ambiguity in discourse caused by erroneous pronouns. Such an approach takes addressees and interpretability into account to explain learner behaviour. Williams (1988) has proposed that learners have two operating principles: a “hyper-clarity” principle (“be as explicit as possible”) for minimal ambiguity, and an economy principle for minimal redundancy. However, it might just as well be argued that learners cause hyperambiguity by using full NPs for maintained referents. By not demoting referents, learners violate a whole range of postulated principles for anaphora resolution: various givenness hierarchies and accessibility scales (Ariel 1990; Chafe 1994; Givón 1984), the Gricean quantity principle (Grice 1975), the minimisation principle (Levinson 1998/1987, 2000), relevance (Sperber & Wilson 1986), principles of recipient design (Sacks & Schegloff 1979), and even the Zipfian principle of least effort (Zipf 1949). It poses obvious problems for interpretation when maintained referents are not formally distinguished from new ones, or when topic-focus status is only indicated by position (Klein & Perdue 1997). For instance, the repeated use of the NP the woman in Figure 2 is in fact an unsuccessful way of resolving coreference since there are several female characters active in the context. Levinson suggests that the normal (read: L1) way of resolving co-reference is by applying the Minimisation principle which can be (carelessly) re-stated as ‘say as little as possible, and interpret as widely as possible’ (‘Less is more’). Learners do the opposite thing by applying a sort of Maximisation principle: ‘say as much as possible, but interpret as narrowly as possible’ (‘More is less’). However, there is both psycholinguistic and interactional evidence suggesting that the outcome of the maximisation principle, over-explicit referring expressions, are difficult to process for addressees. For instance, Cloitre and Bever (1988) showed that appropriately light referential expressions help identify referents better than inappropriately heavy or explicit ones. Pronouns referring to focal referents lead to faster retrieval of information than explicit referring expressions. Similarly, there is ample evidence in the interactional data that native speakers experience problems processing and understanding the over-explicit learner narratives. The native speakers repeatedly engage in negotiations to clarify who actually did what in the story. The effect of over-explicitness in speech is thus ambiguity rather than clarity. Gestural anaphoric linkage offers a potential solution to the communicative problem of hyper-ambiguous over-explicitness. In Figure 2, the deictic gestures allow the addressee to identify the series of NPs la femme as being co-referential since the gestures indicate the same location in space for each expression. The anaphoric gestures allow full visual, unequivocal co-reference to be maintained throughout the sequence. In other words, only by being gesturally explicit is the learner achieving hyper-clarity. Gestural anaphoric linkage can thus serve as a communicative



JB[v.20020404] Prn:26/06/2003; 13:43

F: SIB2612.tex / p.12 (684-740)

 Marianne Gullberg speaker

(a)

(b)

addressee

Figure 3. (a) The addressee (right) pointing to the same locus as indicated by the learner (left). (b) The map of discourse.

trade-off between over-explicit, ambiguous speech, and unambiguous gestures. It is compelling that deictic gestures with NPs often occur in sequences like these where there are several potential antecedents active (cf. Engberg-Pedersen 1993). It is thus tempting to assume that learners disambiguate gesturally what they cannot distinguish in speech. This view of gestural cohesion rests on the assumption that addressees attend to gestures, specifically to (pointing) localising gestures and there is some evidence that they do. When cohesive gestures are deliberately mismatched with speech such that previously established locations are violated in re-activation, addressees find it hard to retell stories coherently (Cassell, McNeill, & McCullough 1999). Moreover, addressees themselves refer to the discourse maps established in interactional space, pointing back to loci established by the learner when negotiating a referent (Gullberg 1998). In doing so, they respect the absolute topological quality of the discourse map and the locations set up in the learners’ space, and indicate exactly the same loci as indicated by the learners instead of setting up their own. The native speaker in Figure 3a points to the same location in space as the learner, thus referring to joint gesture and discourse space (Figure 3b). This communicative, addressee-oriented account of bi-modal anaphora is intuitively attractive. However, it does not fully account for the link between NPs and gestures, nor for the fact that speakers, native and non-native alike, localise referents at their introduction. Moreover, it is not unequivocally clear that learners deliberately externalise and point to discourse for the benefit of the addressee in order to reduce ambiguity and allow for joint solutions (Gullberg 1998; Wilkes-Gibbs 1995). The problem here is well-known from the field of communication strategies: how intentional and/or aware does behaviour have to be in order to qualify as a

JB[v.20020404] Prn:26/06/2003; 13:43

F: SIB2612.tex / p.13 (740-785)

Gestures, referents, and anaphoric linkage in learner varieties 

communication strategy (Gullberg 1998; Kasper & Kellerman 1997)? The localising gestures are clearly performed with a high degree of automaticity and little awareness. When asked if they remember performing these gestures, learners invariably say no – even when the loci and the referents associated with them have been the subject of extensive negotiation. As for other types of communication strategies, learners may ‘subconsciously’ perform gestures in contexts of potential ambiguity much in the same way as they can circumvent other communicative problems before these become manifest. In contrast to what is the case for many spoken strategies, the potentially intentional disambiguating, communicatively driven, effect of gestural anaphoric linkage can (and should) be tested empirically by manipulating addressees’ visual access to learners’ anaphoric gestures. Such a design will tell us to what extent learners actually consider their addressees when deploying anaphoric devices in speech and gesture in conjunction.

. Is it a reflection of learners’ speech planning? An alternative and more speculative view of anaphoric gestures considers the speaker’s own perspective on over-marking in speech and gesture. In such an approach, gestures are not (solely) seen as interactive communicative solutions, but rather as a reflection of speakers’ (and learners’) cognitive efforts to construct utterances, and perhaps even global discourse structure beyond the single utterance level. Under this view, gestural anaphoric linkage could afford a new window into L2 speech planning and processing. The field of gesture studies has recently seen the development of several theories assigning a role to gesture in speech production processes. McNeill’s theory of gesture production, the so-called Growth Point theory (McNeill & Duncan 2000), is particularly interesting for the issue of discourse construction. The growth point in an utterance is similar to Vygotsky’s ‘psychological predicate’, i.e. it represents a significant departure from what precedes in an immediate context, and represents the new “idea”. The growth point is materialised in gesture and speech simultaneously – in linear analytic form in speech, and in global synthetic form in gestures, each modality contributing that part of the information for which it is best suited. Since gestures reflect a new idea, they occur either with predications or with referential expressions in focus, as seen above (Levy & McNeill 1992). This account indirectly explains why there are no gestures with pronouns. Since pronouns inherently express presupposed material that is not part of the growth point, they should not (and do not) receive gestural expression. Another potential consequence of this view is that the units in speech that cooccur with gestures reflect units planned for speech execution. I.e. if a gesture re-

JB[v.20020404] Prn:26/06/2003; 13:43

F: SIB2612.tex / p.14 (785-840)

 Marianne Gullberg

flects a growth point or new material to be expressed, then it might also be seen as a reflection of a planning unit. What does this mean for early L2 acquisition? Learner speech is typically non-fluent and proceeds in small units interrupted by frequent pauses, sometimes between every unit in an utterance. Similarly, several observers have noted that learners at such stages of acquisition typically produce gestures on almost every argument of an utterance (Gullberg 1998; Nobe 1993). The learner in (4) produces a gesture on the predication, another one on the referential expression in direct object position, as well as one on the oblique object. (4) hon eh + [ger] hon ger eh [pa papper] papper eh [till apotekarin] she uh + [gives] she gives uh [pa paper] paper uh [to the pharmacist] If gestures mark growth points or idea units, then more idea units are needed to express all events at early stages of discourse production in L2 as in (4). Gestures would then be a reflection of learners’ L2 speech planning proceeding by smaller units (Nobe 1993). With increasing fluency and automaticity in speech production (lexical retrieval, grammatical encoding), learners’ gestures package more meaning units into one single gesture (e.g. the action of giving and the paper in one gesture). Returning to referential movement, a gesture on a maintained referent expressed as an NP in topic could then be taken to mean that this entity has been planned as a separate unit, much as if it were a new idea. This line of reasoning opens up a range of new questions pertaining to how a processing approach could account for the use of full NPs in speech, and the late acquisition of pronouns; to whether and how gestures and externalisation of discourse help to reduce learners’ cognitive load. These issues remain to be explored both in the field of gesture studies and in SLA research.

. Final remarks Both communicative and cognitive mechanisms are likely to play an important role in the production of over-explicit anaphoric linkage in speech and gesture. Although much remains to be done in this area, especially with respect to the cognitive, speaker-internal perspective, it is hopefully clear by now that gesture analysis can offer important insights into both aspects of anaphoric linkage. In addition, gestural cohesion may also shed new light on stabilisation vs. learners’ continued development. Anaphoric gestures offer a tempting explanation for why learners ‘can’ stabilise at levels where their speech alone is not communicatively efficient. As shown in this paper, the combination of over-explicit anaphora in both modalities provides learners with greater communicative efficiency then they would have in speech alone. The assumption that communicative efficiency prompts stabili-

JB[v.20020404] Prn:26/06/2003; 13:43

F: SIB2612.tex / p.15 (840-888)

Gestures, referents, and anaphoric linkage in learner varieties 

sation is not new. The classical description of the communicatively efficient but formally unsuccessful learner Wes (Schmidt 1983) certainly reinforced this conclusion. However, it is not a foregone conclusion that communicative efficiency is harmful to continued formal acquisition. Communicative efficiency may also be seen as an important contributing factor in the development of language, namely in the sense that it improves learners’ possibilities to produce “comprehensible output” (Swain 1985). Comprehensible output has been suggested as one of the driving forces in language development. If the view is that you have to talk to learn how to talk (Bruner 1990), then anaphoric gestures certainly help you do this in interaction, whether you see them as communicatively or processing driven. In this sense they may actually contribute to the formal development of linguistic means for anaphoric linkage.

Notes . These data were collected within the European Science Foundation project Second Language Acquisition by Adult Immigrants by the teams of Universiteit Brabant, Tilburg, the Netherlands, and Göteborgs Universitet, Göteborg, Sweden. The data were kindly made available to me by the Central Data Archives of this project, located at the Max Planck Institute for Psycholinguistics, Nijmegen, the Netherlands, and by the Dept. of Linguistics, Göteborgs Universitet, Sweden. . The term localisation is borrowed from Sign Language where a similar phenomenon is part of the linguistic, grammatical system. Nominal signs are localised or associated with spatial loci at their introduction. Pronominal signs directed towards these loci are interpreted as co-referential with the nominal. Signs for verbal components are also oriented with respect to these anchor points such that localisation is part of the grammatical means to achieve agreement (Engberg-Pedersen 1993; Liddell 1996). . See for instance Levinson (1998/1987: 598) on the extensive usage of zero anaphora in Guugu Yimithirr: “[. . . ] it is hardly surprising that a language with zero anaphora and no verbal agreement would find an ancillary channel of gestural information very useful indeed [. . . ].” It is important to note, however, that this disambiguation is not a function of gestures co-occurring with zero anaphora. Gestures that can disambiguate zero anaphora are instead iconic gestures that appear with the predication of an utterance, rather than with the referring expression. Iconic gestures can indirectly express agent or patient by being oriented towards previously established locations in space much as in Sign Language. And indeed, in the GY data pointing gestures co-occur only with NPs, demonstratives (±N), or location adverbials, similarly to the learner data under consideration, and not with zero anaphora or indeed with pronouns themselves. Note also that the disambiguating function of gestures with respect to expressions like demonstratives is not under discussion here, since this is an exophoric use of gestures, not an endo- or anaphoric one.

JB[v.20020404] Prn:26/06/2003; 13:43

F: SIB2612.tex / p.16 (888-991)

 Marianne Gullberg

References Ahrenholz, B. (1998). Modalität und Diskurs: Instruktionen auf deutsch und italienisch. Eine Untersuchung zum Zweitspracherwerb und zur Textlinguistik. Tübingen: Stauffenburg. Ariel, M. (1990). Accessing noun phrase antecedents. New York: Routledge. Bruner, J. (1990). Acts of meaning. Cambridge, MA: Harvard University Press. Carroll, M., Murcia-Serra, J., Watorek, M., & Bendiscoli, A. (2000). The relevance of information organization to second language acquisition studies: The descriptive discourse of advanced adult learners of German. Studies in Second Language Acquisition, 22(3), 441–466. Cassell, J., McNeill, D., & McCullough, K.-E. (1999). Speech-gesture mismatches: Evidence for one underlying representation of linguistic and nonlinguistic information. Pragmatics & Cognition, 7(1), 1–33. Chafe, W. (1994). Discourse, consciousness, and time. Chicago: University of Chicago Press. Cloitre, M. & Bever, T. G. (1988). Linguistic anaphors, levels of representation and discourse. Language and Cognitive Processes, 3(4), 293–322. Engberg-Pedersen, E. (1993). Space in Danish Sign Language. Hamburg: Signum. Extra, G., Strömqvist, S., & Broeder, P. (1988). Pronominal reference to persons in adult second language acquisition. In P. Broeder, G. Extra, R. v. Hout, S. Strömqvist, & K. Voionmaa (Eds.), Processes in the developing lexicon (pp. 86–113). Strasbourg, Tilburg, Göteborg: European Science Foundation. Fox, B. (1987). Discourse structure and anaphora. Cambridge: Cambridge University Press. Givón, T. (1984). Universals of discourse structure and second language acquisition. In W. E. Rutherford (Ed.), Language universals and second language acquisition (pp. 109–136). Amsterdam: Benjamins. Grice, H. P. (1975). Logic and conversation. In P. Cole & J. L. Morgan (Eds.), Syntax and semantics: Speech acts (pp. 41–58). New York: Academic Press. Gullberg, M. (1998). Gesture as a communication strategy in second language discourse. A study of learners of French and Swedish. Lund: Lund University Press. Hendriks, H. (2000). The acquisition of topic in Chinese L1 and French L1 and L2. Studies in Second Language Acquisition, 22(3), 369–397. Jin, H. G. (1994). Topic-prominence and subject prominence in L2 acquisition: Evidence of English-to-Chinese typological transfer. Language Learning, 44, 101–122. Kasper, G. & Kellerman, E. (1997). Introduction: approaches to communication strategies. In G. Kasper & E. Kellerman (Eds.), Communication strategies: Psycholinguistic and sociolinguistic perspectives (pp. 1–13). London: Longman. Kendon, A. (1972). Some relationships between body motion and speech: An analysis of an example. In A. W. Siegman & B. Pope (Eds.), Studies in dyadic communication (pp. 177–210). New York: Pergamon. Kendon, A. (1993). Human gesture. In K. R. Gibson & T. Ingold (Eds.), Tools, language and cognition in human evolution (pp. 43–62). Cambridge: Cambridge University Press. Klein, W. & Perdue, C. (1997). The basic variety (or: Couldn’t natural languages be much simpler?). Second Language Research, 13(4), 301–347. Lambrecht, K. (1994). Information structure and sentence form. Topic, focus, and the mental representation of discourse referents. Cambridge: Cambridge University Press.

JB[v.20020404] Prn:26/06/2003; 13:43

F: SIB2612.tex / p.17 (991-1090)

Gestures, referents, and anaphoric linkage in learner varieties 

Levinson, S. C. (1998/1987). Minimization and conversational inference. In A. Kasher (Ed.), Presupposition, implicature and indirect speech acts (pp. 545–614). London: Routledge. Levinson, S. C. (2000). Presumptive meanings. The theory of generalized conversational implicature. Cambridge, MA: MIT Press. Levy, E. T. & McNeill, D. (1992). Speech, gesture, and discourse. Discourse Processes, 15(3), 277–301. Liddell, S. K. (1996). Spatial representations in discourse: comparing spoken and signed language. Lingua, 98(1–3), 145–167. Marslen-Wilson, W., Levy, E., & Komisarjevsky, Tyler, L. (1982). Producing interpretable discourse: The establishment and maintenance of reference. In R. J. Jarvella & W. Klein (Eds.), Language, place, and action. Studies in deixis and related topics (pp. 339–378). Chichester: John Wiley. McNeill, D. (1986). Iconic gestures in children and adults. Semiotica, 62(1/2), 107–128. McNeill, D. (1992). Hand and mind. What the hands reveal about thought. Chicago: Chicago University Press. McNeill, D. (2000). Catchments and contexts: nonmodular factors in speech and gesture production. In D. McNeill (Ed.), Language and gesture (pp. 312–238). Cambridge: Cambridge University Press. McNeill, D. & Duncan, S. D. (2000). Growth points in thinking-for-speaking. In D. McNeill (Ed.), Language and gesture (pp. 141–161). Cambridge: Cambridge University Press. McNeill, D. & Levy, E. T. (1993). Cohesion and gesture. Discourse Processes, 16(4), 363–386. McNeill, D. & Pedelty, L. L. (1995). Right brain and gesture. In K. Emmorey & J. S. Reilly (Eds.), Language, gesture, and space (pp. 63–85). Hillsdale, NJ: Erlbaum. Muñoz, C. (1995). Markedness and the acquisition of referential forms. The case of zero anaphora. Studies in Second Language Acquisition, 17(4), 517–527. Muñoz, C. (2000). The over-explication of personal reference in L2 narratives. Paper presented at the Euresco Conference ‘The structure of learner language’, San Feliu de Guixols, October 7–12, 2000. Nobe, S. (1993). Cognitive processes of speaking and gesturing: A comparison between first language speakers and foreign language speakers. Unpublished Masters thesis, Dept. of Psychology, University of Chicago. Pekarek, S. (1999). Linguistic forms and social interaction: Why do we specify referents more than is necessary for their identification? In J. Verschueren (Ed.), Pragmatics in 1998. Selected papers from the 6th International Pragmatics Conference (pp. 427–448). Antwerpen: International Pragmatics Association. Sacks, H. & Schegloff, E. A. (1979). Two preferences in the organization of reference to persons in conversation and their interaction. In G. Psathas (Ed.), Everyday language: Studies in ethnomethodology (pp. 15–21). New York: Irvington. Schmidt, R. W. (1983). Interaction, acculturation, and the acquisition of communicative competence: A case study of an adult. In N. Wolfson & E. Judd (Eds.), Sociolinguistics and language acquisition (pp. 137–174). Rowley, MA: Newbury House. Sperber, D. & Wilson, D. (1986). Relevance. Communication and cognition. Oxford: Blackwell. Strömqvist, S. & Day, D. (1993). On the development of narrative structure in child L1 and adult L2 acquisition. Applied Psycholinguistics, 13(2), 135–158.

JB[v.20020404] Prn:26/06/2003; 13:43

F: SIB2612.tex / p.18 (1090-1114)

 Marianne Gullberg

Swain, M. (1985). Communicative competence. Some roles of comprehensible input and comprehensible output in its development. In S. M. Gass & C. G. Madden (Eds.), Input in second language acquisition (pp. 235–253). Rowley, MA: Newbury House. von Stutterheim, C. & Klein, W. (1989). Referential movement in descriptive and narrative discourse. In R. Dietrich & C. F. Grauman (Eds.), Language processing in social context (pp. 39–76). Amsterdam: North-Holland. Wilkes-Gibbs, D. (1995). Coherence in collaboration: Some examples from conversation. In M. A. Gernsbacher & T. Givón (Eds.), Coherence in spontaneous text (pp. 239–267). Amsterdam: Benjamins. Williams, J. (1988). Zero anaphora in second language acquisition. Studies in Second Language Acquisition, 10, 339–370. Zipf, G. K. (1949). Human behavior and the principle of least effort: An introduction to human ecology. New York: Hafner.

The development of anaphoric means to refer to space and entities in the acquisition of French by Polish learners Marzena Watorek

.

Introduction

In this chapter, we present the development of the linguistic means used to construct anaphoric relations in the domains of space and entities in descriptive discourse. In a cross-sectional and cross-linguistic study, we analyse the development of linguistic means in these two referential domains for two groups of learners: intermediate and advanced Polish learners of French, and of two control groups: French and Polish native speakers. This latter analysis allows us better to evaluate the influence that different ways of organising information in French and Polish, in this kind of text in particular, have on the processes involved in constructing discourse in the target language. Our main questions are: A. How does the use of locative expressions change across the two groups of learners? B. More precisely: what are the spatial relations expressed, do they complexify with increasing proficiency? C. How does anaphoric linkage develop across the two groups?

. Informants and data Our database consists of static spatial descriptions by 8 intermediate learners who are second year students in the Department of Romance Languages at a Polish university (henceforth “group I”), and 9 advanced learners in their fourth year of studies in this same department (henceforth “group II”).1 These students are future specialists of French (cf. Bartning’s apprenant avancé d’instruction élévée – ‘advanced learners with a high degree of education’; 1997) and were selected on the

 Marzena Watorek

basis of a socio-linguistic questionnaire to ensure that their acquisitional histories are comparable. The communicative task analysed is a spatial description. All learners and native speakers (10 French and 10 Polish native speakers) were asked to describe a poster of a town to an unknown interlocutor who was unfamiliar with the poster.2 The interlocutor was then asked to draw a picture based on the subject’s description. This type of description (cf. Watorek 1996a, 1998) can be defined as a response to the question-type (or quaestio, Klein & von Stutterheim 1991): “where is what (in the picture)?”. During the process of constructing discourse, this quaestio can be realised in two ways (cf. Klein & von Stutterheim’s “referential movement”, 1991): the more frequent realisation states the quaestio as, “what is there in L?” where L corresponds to the total space of the poster; the less frequent realisation asks, “where is this X?” where X corresponds to the entity already localised a first time. Hence, the utterances of the main structure in these texts can have two topic/focus organisations. In the case of the more frequent realisation of the quaestio, the information about space is in the topic, while the entity to localise is the focus of the utterance. In the other, less frequent, realisation of the quaestio the entity is in the topic, whereas the localising space is in the focus of the utterance. We will show that the distinction between these two realisations of the global quaestio is important for anaphoric linkage in that discourse cohesion is greater when an entity already localised for the first time, and is maintained in the topic of the following utterance for a more precise localisation. The global quaestio “where is what in the picture?” constrains the informational content of the main structure utterances. Thus, every utterance of the main structure express the spatial relation between a reference entity defining the spatial context (relatum), and an entity to be localised (theme) in the global space being described, the poster.3 This spatial relation can be topological or projective. In the topological relations of inclusion, exclusion, neighbouring, etc., the intrinsic features of theme and relatum are essential because the spatial spans are delimited by these entities. For projective deictic relations, it is the orthogonal axis system (lateral/sagittal/horizontal) that allows the spatial localisation, thus the features of theme and relatum are less important since the origo of the system is the speaker. However, if the projective relations are expressed by a system of axes of an oriented relatum-object (intrinsic projective relations – a car has a “left” and a “right”, for example) then the features of entities can be important for localisation. Speakers use the spatial concepts and frames of reference to interrelate spatial information in a coherent text, which reflects different ways of conceptualising the configuration under description. Carroll and von Stutterheim (1997) analyse the patterns of organisation used in the descriptions of German and English native speakers as compared to English learners of German. They find two basic pat-

The development of anaphoric means

terns of organisation of spatial information: the global frame of reference and the point-by-point frame of reference. We illustrate these patterns with examples from our own data. In the global frame, the speakers choose a complex structure, e.g., a system of coordinate axes (cf. example (1a)) or a specific sub-space of the global relatum (cf. example (1b)) projected onto the entity under description (i.e., the poster) to define sections and corresponding regions of space to localise the themes. (1) a.

a gauche il y a un bâtiment jaune on the left there is a yellow building b. au premier plan de l’affiche il y a une rue in the foreground of the poster there is a street

In the point-by-point frame, the information is expressed in terms of the individual objects that make up the entity under description. The themes are localised in the region associated with single reference objects represented in the picture (cf. example (2)). (2) devant la fontaine il y a un arrêt de bus in front of the fountain there is a bus stop The typical informational content in this kind of discourse corresponds to the static localisation that can be paraphrased as: “in the region of relatum X there is a theme Y” where the type of relatum chosen by the speaker determines one or the other of the two frames of reference. This information is encoded in the utterance by the general phrasal pattern PP/ADV + V + NP, common in both French and Polish, the languages considered in our study.4 However, some differences between Polish and French selectively influence the structure of the discourse produced by the respective native speakers. We also observe that the learners’ production is constrained by differences in the structure of their variety at the intermediate and advanced level of acquisition. In French, the finite verb (Vfin) in this syntactic pattern corresponds to the existential verb form il y a. In Polish where existential forms do not exist, we find locative verbs like by´c ‘to be’ or stac ‘to stand’, lezec ‘to lie’ etc. This difference has consequences for anaphoric linkage, as we shall see below. The type of locative expressions available in French and in Polish5 also contribute to anaphoric linkage and influences the choice of the frame of reference (global or point-by-point) as well as the choice of the referential domain for marking discourse cohesion. Thus, the structure and the semantics of the available expressions constrain discourse construction in different ways as already shown by Carroll and von Stutterheim (1997) and Watorek (1996b).



 Marzena Watorek

. The results We present the results of our study in two steps. First, we report the type of spatial concepts expressed by the learners of the two groups (question [B] of the introduction) and the linguistic means used by each group. We conclude that the complexification of the spatial concepts expressed in the learners’ descriptions (cf. Figure 1 in §3.1.1) depends on the linguistic means that are most easily mobilised (cf. Watorek 1996b) in the learner language of each group. Second, we present the analysis of the locative expressions. We show firstly how learners choose one or the other of the referential domains (entities or space) to mark discourse cohesion. Secondly, the analysis of the structure of the NP-relatum included in the locative expression points to a developmental path for anaphoric linkage. The developmental path identified shows similarities with those found in longitudinal studies of first and second language acquisition. We will return to this in the conclusion.

. Development of spatial concepts and anaphoric linkage In this section we will first discuss the complexification of spatial concepts expressed in learner and native speaker descriptions (§3.1.1). We will then demonstrate that discourse coherence and anaphoric linkage depend to some extend on the phrasal patterns that speakers choose to encode spatial information (§3.1.2).

.. Complexification of spatial concepts Our analysis shows that the type of spatial concepts expressed in the descriptions changes between group I and group II. In fact, we observe that the learners of group II use locative expressions which encode more complex spatial concepts. Figure 1 shows this evolution. Topological relations Topological simple ‘at place’

>

>

Topological complex

Projective relations point-by-point: intrinsic global frame: deictic

> >

deictic intrinsic

-inclusion -neighbouring -boundary

global frame expressed as topological relations

>

global frame expressed as projective relations

Figure 1. Type of spatial concepts expressed by the two groups of learners

The development of anaphoric means 

Generally, the intermediate learners of group I use proportionately more topological relations than the advanced learners of group II. We see this in Table 1. Another fact confirms this tendency. The global frame of reference is conceptualised by the learners of group I with topological relations, whereas the learners of group II use more projective relations (cf. Table 2). It becomes clear from Table 2 that the native speakers of French and Polish tend to present more projective than topological relations in the global frame. This is not the case in the descriptions of the learners of group I. In other words, we find more expressions like sur l’affiche ‘in the poster’, au centre de l’affiche ‘in the centre of the poster’ in the descriptions of group I, whereas in the descriptions of the natives and the learners of group II we find more expressions like au premier plan de l’affiche ‘in the foreground of the poster’, devant ‘in the front’ for the sagittal axis, à gauche/à droite de l’affiche ‘on the left/right of the poster’ for the lateral axis and en haut/en bas de l’affiche ‘at the top/bottom of the poster’ for the vertical axis. Table 1. Percentages of locative expressions encoding topological and projective relations Topological

Projective

POL > FR (I) total 62

53 = 85.5%

9 = 14.5%

POL > FR (II) total 208

136 = 65.4%

72 = 34.6%

FRENCH total 485

258 = 53.2%

227 = 46.8%

POLISH total 655

374 = 57.1%

281 = 42.9%

Table 2. Locative expressions encoding topological and projective relations used in the global frame of reference Global frame Topological relations

Projective relations

POL > FR (I) total 9

6 = 66.6%

3 = 33.3%

POL > FR (II) total 47

11 = 23.4%

36 = 76.6%

FRENCH total 125

14 = 11.2%

111 = 88.8%

POLISH total 181

18 = 9.9%

163 = 90.1%

 Marzena Watorek

A comparison between these two groups of learners also shows that simple topological relations like ‘at place’ are expressed proportionately more frequently by the learners of group I than the more complex topological relations of inclusion and neighbouring. The topological relation of boundary is only expressed by group II (see Table 3). In fact, the relation ‘at place’ is cognitively easier. It demands only a simple relation between a theme and a relatum without any constraint on the intrinsic features of either object. In the following example, The relatum is expressed, then the theme, and the precise relation between them is left vague. (3) l’arrêt de bus une dame the bus stop a lady In the relation of inclusion, on the contrary, the relatum must be an entity with the features of containment, and the preposition dans ‘in’ reflects this feature. (4) dans la cage il y a un oiseau in the cage there is a bird

Table 3. The diversification of the expression of topological relations in the place of

inclusion

neighbouring

boundary

other

POL > FR (I) total 53

13 = 24.5%

14 = 26.4%

8 = 15.1%

0 =0%

18 = 34%

POL > FR (II) total 136

34 = 25%

48 = 35.3%

24 = 17.6%

6 = 4.4%

24 = 17.7%

Table 4. Locative expressions encoding intrinsic and deictic projective relations used in the global frame versus the point-by-point frame Projective relations Intrinsic Deictic global frame point-by-point frame global frame point-by-point frame POL > FR (I) total 9

0 = 0%

6 = 66.6%

3 = 33.3%

0 =0%

POL > FR (II) 19 total 72 = 26.4%

24 = 33.3%

17 = 23.6%

12 = 16.6%

POLISH total 281

75 = 26.7%

78 = 27.8%

88 = 31.3%

40 = 14.2%

FRENCH total 227

56 = 24.7%

85 = 37.4%

55 = 24.2%

31 = 13.7%

The development of anaphoric means 

We also observe other tendencies. When the speaker organises spatial information in the point-by-point frame, projective intrinsic relations are expressed before deictic relations. The contrary is observed when the speaker organises spatial information in the global frame (cf. Table 4). In the point-by-point frame, the intrinsic projective relations are expressed more in group I productions than the deictic ones which are more frequent in the descriptions of group II. This order of expression of the spatial projective relations is connected to the possibility of conflicts between the position of the speaker and the origo of the relatum, in cases where the relatum is oriented. (5) à droite de la voiture bleue il y a un homme en vélo to the right of the blue car there is a man on a bike In example (5), the car and the speaker have their “right” sides, which do not necessarily coincide. In this case, the speaker can present this scene in terms of the deictic projective relations if he bases the localisation on his own perspective. On the contrary, if the localisation is based on the object-relatum position (car), the speaker expresses this scene in terms of the intrinsic relations. Where the relatum has neither lateral nor sagittal orientation, only the deictic interpretation of the projective relations is possible as with the relatum tree in example (6). (6) à gauche de l’arbre il y a un homme qui donne à manger aux pigeons to the left of the tree there is a man who gives to eat to the pigeons Hence, when the learners of group II start to express the projective deictic relations, they only choose non-oriented entities for the relatum. Thus, if the learner expresses a projective relation in the lateral or sagittal axis, he selects non-oriented relata, such as river, bridge, tree or garden. Concerning the global frame, the projective relations are expressed first as deictic and then as intrinsic (see Table 4). In contrast to the point-by-point frame, in the global frame the relatum-poster is only oriented in the vertical axis, which corresponds to the speaker’s position.6

..

The influence of the learners’ linguistic means on the spatial concepts expressed We find tree basic phrasal patterns which characterise the structure of the learnervarieties of both groups: (A) PP/ADV + Vexistential + NP (B) Vexistential + NP (C) Vexistential + aussi + NP

 Marzena Watorek

The choice of phrasal patterns changes with the competence level of the learners and influences the complexity of the spatial relations expressed and the degree of cohesion produced by each group of learners. The learners of group I use pattern (B) in which the PP/ADV component is more frequently left unexpressed than by the learners in group II. The ellipsis of the PP/ADV is frequently correlated with the presence of the additive scope particle aussi ‘also’ as in pattern (C): il y a + aussi + SN. In their descriptions, the advanced learners of group II use the complete phrasal pattern (A) more frequently: PP/ADV + il y a + NP. Let us now look at how the different uses of linguistic means by the learners of the two groups influences the organisation of spatial information in the analysed texts.

Pattern (B): il y a + NP The relatum and the spatial span delimited by it is more often left implicit in the discourse produced by the learners of group I than that of group II. We observe that the learners of group I more frequently omit the locative expression (PP/ADV) encoding the relatum and the space delimited by it. Example (7) shows the first utterances of a description of a learner from group I. (7) MAR, Polish > French (group I) 1 il y a + il y a un marché there is + there is a market 2 une petite rivière le pont a little river the bridge 3 il y a + il y a des arbres there is + there are trees. 4 c’est intéressant que / it’s interesting that/ 5 ah on peut voir des des montagnes ici ah we can see mountains here 6 et quoi d’autre ? + il n’y a pas beaucoup de monde and what else ? + there aren’t many people 7 dans la rue il y a des enfants qui jouent au ballon in the street there are children that are playing ball We can see in this example that the dominant phrasal pattern is il y a + NP, where the NP refers to the theme. A locative expression referring to the relatum and the space delimited by it, pattern (A), is made explicit only in utterance 7. In Table 5, we clearly see that the omission of the PP/ADV is proportionately less frequent in group II and in the texts of native speakers of both languages.

The development of anaphoric means 

Table 5. Main structure utterances with or without locative expression referring to the relatum implicit spatial expression

explicite spatial expression

POL > FR (I) total 174

107 = 61,5%

67 = 38,5%

POL > FR (II) total 340

121 = 35,6%

219 = 64,4%

POLISH total 677

90 = 13,3%

587 = 86,7%

FRENCH total 623

193 = 31%

430 = 69%

Table 6. Percentages of locative expressions referring to the global frame or point-bypoint frame Global

Point-by-point

POL > FR (I) total 62

9 = 14.5%

53 = 85.5%

POL > FR (II) total 208

47 = 22.6%

161 = 77.4%

FRENCH total 485

125 =25.7%

360 =74.3%

POLISH total 655

181 = 27.6%

474 = 72.4%

Furthermore, we note that in group I, the explicit mention of the relatum is more frequent when the learners express the spatial relation with the point-bypoint frame of reference (example (7)). On the other hand, explicit mention of the global poster-relatum, or its parts, is less frequent in the productions of group I, as seen in Table 6. The explicit mention of the relatum in the global frame clearly becomes more frequent in group II. This means that the learners of group I consider the global relatum as presupposed with regard to the quaestio. They construct their descriptions based on the spatial relation between the themes and the relata corresponding to individual objects. It goes without saying that for these learners, this theme – relatum relation is within the global space being described. The phrasal pattern (B) V existential + NP is frequent in the French data from native speakers and learners (cf. Table 7). But it becomes particularly productive in the descriptions of group I. This pattern then evolves in the descriptions of group II towards the pattern where PP/ADV is more often explicit (cf. Table 5). In addition,

 Marzena Watorek

Table 7. Percentages of the different verb types attested in the descriptions locative verbs

existential verbs

other

POL > FR (I) total 174

11 =6,3%

129 = 74,1%

34 = 19,6%

POL > FR (II) total 340

29 = 8,5%

237 = 69,7%

74 = 21,8%

POLISH total 679

362 = 53,3%

118 = 17,3%

199 = 29,3%

FRENCH total 605

44 = 7,3%

358 = 59,1%

203 = 33,6%

we observe a diversification of the verb types used. Table 7 depicts the percentage of the different types of verbs found in the data. Existential verbs are dominant in the descriptions of the two groups of learners and more frequent than in French native speaker productions. However, we observe a decrease of this verb type in the discourse of group II, bringing them closer to the native speakers of French. The existential verb is so productive in this kind of discourse in French L1 and L2 (cf. Watorek 1996a, 1998) because it allows the NP that refers to the theme in focus to be at the end of the utterance (“focus expression last”). Most main structure utterances “answer” the quaestio “what is there in L?”: the information regarding the theme is in focus and the spatial span (L) is presupposed. As for the locative verbs, they are dominant in the descriptions of the native speakers of Polish and rare in the descriptions of the French natives. As we said in 1 above Polish has no true existential verb form like French, but it makes use of the locative form byc ‘to be’ and other locative verbs that carry the semantics of to be or to exist. These verbs also permit an NP in the post-verbal position, again correlating with an informational structure where the focus information is generally placed in final position. In French this category of verb scarcely exists and the use of the locational verbs être’ ‘to be’ and se trouver ‘to find oneself ’ is restricted by the “identifiability constraint” (cf. Carroll et al. 2000): only a theme already identified in the context can be localised with this form. Utterances such as au centre de la place est une fontaine ‘in the centre of the square is a fountain’ are less natural in French than in Polish (or German). The copula is only used when an already identified theme – therefore referred to by a definite NP – is in pre-verbal position: la/cette fontaine est sur la place ‘the fountain is in the square’. Thus, the locative is used in French only when the theme is in the topic. In this case the utterance is an answer to the less frequent realisation of the quaestio “where is this X?”, where X corresponds to the identified theme. Like the native speakers of French, the learners of both groups employ the locatives only when the utterance answers “where is

The development of anaphoric means 

this X?”. However, we observe an increase in use of the locative “être” from group I to group II. This means that in the descriptions of the advanced learners the second realisation of the quaestio is more frequent. This increase is significant for betweenutterance anaphoric linkage, which is stronger. Actually, the presence of utterances which answer, “where is this X?” implies the immediate maintenance of reference and a second localisation of the theme, which was introduced and localised for the first time in the preceding utterance. Look at the following examples. (8) AGK, Polish > French (group II) 1 je vois une rue I see a street 2 qui est en bas du dessin that is at the bottom of the picture 3 dans cette rue il y a beaucoup de bâtiments in this street there are many buildings The theme which was introduced and localised in utterance 1 is then maintained by the pronoun qui ‘that’ and localised more precisely. The description in example (8) is more cohesive than that in example (9), taken from group I. (9) AGA, Polish > French (group I) 1 je vois une rue I see a street 2 il y a beaucoup de bâtiments there are many buildings In example (9), the theme in utterance 2 bâtiments ‘buildings’ is localised ‘at the place’ delimited by the inferred relatum for 2 (‘street’). This relatum is also implicitly localised in the global space of the poster in utterance 1. From this demonstration, we conclude that the learners of both groups have acquired the identifiability constraint of the French locative. The use of the pattern definite NP + est + PP permits the construction of discourse with a stronger anaphoric linkage, which becomes more important in the descriptions of the advanced learners of group II. We note that if the relatum is more frequently implicit at first, and if the spatial relations expressed are more frequently simple as in ‘at place’, the phrasal pattern (B) V existential + NP is minimally sufficient. We can also say that since this phrasal pattern is more easily mobilisable in the context of the spatial description the speakers organise the information in the discourse in such a way that the information can be encoded with the help of this minimal pattern (B). Hence, speakers must select an entity which is salient enough to take on the role of relatum for a large number of themes. Thus, the simple spatial topological relation ‘at place of a

 Marzena Watorek

relatum’ is expressed; the relatum can be kept constant and consequently left implicit. The learners more than natives, and the learners of group I more than the learners of group II, tend to do this. This fact reflects group I’s more restricted spatial repertoire: the spatial configuration being described is more often conceptualised in topological terms in group I productions than in the productions of group II.

Pattern (C): V existential + AUSSI + NP As noted in the previous paragraphs, the learners of group I carried out the descriptive task in a minimal manner, constructing discourse in which cohesion is relatively weakly marked, as can also be seen in the following example: (10) ANG, Polish > French (group I) 1 je vois une grande rue I see a big street 2 c’est une rue dans une grande ville It is a street in a big town 3 je vois beaucoup de gens I see many people 4 il y a des gens qui se promènent there are people who walk 5 il y a une femme qui vend des magazines there is a lady who sells magazines 6 il y a des gens des enfants qui jouent qui s’amusent there are people children who play who enjoy themselves 7 il y a aussi des gens qui font du vélo there are also people who ride a bike 8 il y a un restaurant there is a restaurant 9 il y a aussi deux trois rues there are also two three streets 10 il y a aussi une rivière there is also a river In this example the relatively weak anaphoric linkage is strengthened by the additive scope particle aussi ‘also’ (see utterances 7, 9, 10). As was demonstrated by Watorek (1996a, b), Watorek and Perdue (1999) and Watorek and Dimroth (in press), additive scope particles play a significant role as anaphoric means in the process of discourse construction. These particles reinforce the cohesion of the discourse in that they indicate explicitly that some information from the context is maintained. In example (10), the entity une grande rue ‘a big street’ is intro-

The development of anaphoric means 

Table 8. Frequency of “aussi” in main structure utterances with implicit and explicit relatum aussi + implicit relatum

aussi + explicit relatum

POL > FR (I) total 22

17 = 77,3%

5 = 22,7%

POL > FR (II) total 36

25 = 69,4%

11 = 30,6%

FRENCH total 37

14 = 38%

23 = 62%

POLISH total 37

16 =43,2%

21 =56,8%

duced in utterance 1 and implicitly maintained in the following set of utterances as relatum. Aussi ‘also’ in these utterances indicates explicitly that this information corresponding to the relatum is to be maintained. The weaker the anaphoric reference to the relatum, the more aussi is used. This tendency is confirmed in our study as can be seen in Table 8.

Pattern (A): PP/ADV + V existential + NP When the speaker wants to express not only simple topological relations but also complex topological and projective relations, he must use more complex linguistic means to encode them in the discourse. If the space delimited by a relatum is not maintained in its entirety (as in the case of simple topological relations), that is, where it must be subdivided and presented in its various aspects, implicit maintenance of the spatial span is not always possible. Two types of complexification of spatial concepts emerge when the PP/ADV component corresponds to anaphoric locative expressions, implying operations of cohesion in the referential domains of entities and space. First, we observe the diversification of the topological relations presented in Table 3 above. The conceptualisation of the spatial configurations being described as topologically simple relations – “at place” – evolves and diversifies. Thus, the learners of group II more frequently express complex topological relations such as ‘inclusion’ and ‘neighbouring’, and they start to express the topological relation of ‘boundary’. Second, we note that the spatial configurations being described are more frequently presented as projective relations in the productions of group II (see Table 1 above). These projective relations are intrinsic and deictic independently of the type of the frame of reference used (global or point-by-point; cf. Table 4 above). This means that the advanced learners use proportionately more complex locative expressions encoding projective relations. The expression of these concepts seems

 Marzena Watorek

to be more complex because the relation is made between the speaker’s position and the position of the oriented relatum (cf. §3.1.1 above).

. Locative expressions In this task, the locative expression (PP/ADV) is the most important constituent of pattern (A) for the anaphoric linkage. Locative expressions – prepositions, adverbs and complex prepositional phrases – encode the spatial relations between the relatum and the theme, while constructing discourse cohesion. In analysing these linguistic means, we consider two points: I.

the choice of the referential domain of entities versus space used to mark cohesion (cf. §3.2.1); II. the structure of the NP that refers to the relatum and the development of anaphoric linkage (cf. §3.2.2).

.. Entities or space in discourse cohesion According to Carroll et al. (2000) speakers of different languages show preferences for one of the referential domains, that of space or entities, in the process of constructing static spatial descriptions. These preferences depend on the structure of the locative expressions, their semantics, and their use in the discourse context. These authors discuss differences between German and English speakers in terms of the distinction between place and space concepts which reflect the overall perspective taken in information organisation. We also consider these concepts as significant for the analysis of our informants. A place is defined as an area delimited by an object and its outer boundaries. In example (11), utterance 2, the themes are at the place delimited by the object “square”. In this case the spatial span referred to in the locative expression is delimited by the object that is the relatum. This spatial span comes with the physical properties of this object. (11) 1 2

au centre il y a un square in the centre there is a square dans ce square il y a une fontaine et un marchand de fleurs in this square there are a fountain and a seller of flowers

On the contrary, the concept of place does not apply to the area delimited by “devant” and “derrière” in example (12). (12) 1 2

c’est une image d’une ville this is a picture of a town devant il y a une rue principale in the front there is a main street

The development of anaphoric means 

3

derrière il y a des montagnes behind there are mountains

In utterances 2 and 3, it is not an object which defines the area referred to by “devant” (in the front) and “derrière” (behind) but the sagittal axis. Here we are dealing with the space and not the place concept. If the speakers’ organisation of spatial information is based on a place, the domain of entities (i.e., the relatum-object) is crucial for discourse cohesion. On the contrary, when the texts produced are based on the concept of space, the referential domain of space is essential for discourse cohesion. A comparison between Polish and French productions shows interesting differences in the choice of referential domains for anaphoric linkage, the former choosing space while the latter prefer entities. We will illustrate these differences, basing our demonstration on projective relations along the sagittal axis. These relations can be encoded in both languages in spatial expressions of two types: (i) object-neutral and (ii) object-specific expressions. The picture can be divided into sections that can reflect an object-based perspective. The concepts foreground/background are object-based because they are related to inherent features of the picture. French expressions like au premier plan ‘at the foreground’ / en arrière plan ‘at the background’ cannot be used to locate an entity in front of an entity other than the picture/poster itself (p.ex. *au premier plan de la fontaine ‘at the foreground of the fountain’). These are object-specific expressions that refer to the area delimited by an object and its boundaries. Their use gives more importance to the concept of place. The concepts foreground/background stand in contrast to the more widely applicable concepts of front/back which can apply to a range of spaces. In French, object-neutral expressions like devant ‘in the front’ / derrière ‘behind’ can be used to locate a theme in front of any relatum (any object on a sagittal axis). In this way, their use accords more importance to the concept of space. An analysis of the descriptions in Polish and French by native speakers shows that the Polish speakers employ the object-neutral expressions more frequently than the French speakers do. This is linked to differences in the repertoire of these expressions in the two languages which reveal the following general tendency: the concept of space and the referential domain of space play a more important role in the discourse of the Polish speakers, and the concept of place and the domain of entities are more important in the discourse of the French speakers. Tables 9 and 10 show the repertoire and the frequency of use of the two types of expressions in the texts of native speakers. We separately present these expressions for the global and the point-by-point frame of reference because this distinction is relevant for our analysis.

 Marzena Watorek

Table 9. Repertoire of native speakers’ locative expressions involving the sagittal axis in the global frame of reference transitive

object-specific intransitive

POLISH

na pierwszym planie rysunku (in the foreground of the image) w glab ˛ rysunku (deep into the image)

na pierwszym planie (in the foreground) na dalszym planie (in the middle distance) w glab/w ˛ gl˛ebi7 (deep inside) w tle (in the background)

Total 86

3 (= 3.5%)

26 (= 30.2%)

FRENCH au premier plan du dessin (in the foreground of the image) au fond du dessin (in the background of the image)

au premier plan (in the foreground) en arrière-plan (in the background) au fond (in the background)

Total 38

22 (= 57.9%)

4 (= 10.5%)

object-neutral transitive intransitive w dali/w oddali (in the distance) dalej (farther away) bli˙zej (closer) w tyle (behind) z tyłu (from behind) z przodu (from in front) na wprost (across from) en face (across from) 0 (= 0%) 57 (= 66.3%) plus loin (further away) devant (in front of ) derrière (behind) en face (across from)

0 (= 0%) 12 (= 31.6%)

In the Polish native speakers’ productions the most frequent object-neutral expressions in the global frame (Table 9) are z przodu/z tyłu ‘in front of/at the back’. In general, these expressions can be employed in an intransitive or a transitive manner. But in this context the Polish speakers use these expressions only in their intransitive form, that is, like adverbs. Thus, they encode the spatial span without reference to the relatum-object. The adverbs are not linked to any specific object or the area of space delimited by its outer boundaries. (13) z tyłu wida´c pasmo gór at the back we see a range of mountains The theme is localised in the space defined by the sagittal axis. It is not possible to identify the relatum-object. Certainly, the poster does not correspond to the relatum-object because z tyłu afisza ‘at the back of the poster’ means out of the pictorial representation under description. In the point-by-point frame of reference (Table 10) where the spatial link is established between singular objects, Polish native speakers use transitive object-

The development of anaphoric means 

Table 10. Repertoire of native speakers’ locative expressions involving the sagittal axis in the point-by-point frame of reference transitive

object-specific intransitive

transitive

object-neutral intransitive

FRENCH devant (in front of ) 20 derrière (behind sth) 5 en face de (across from) 5 Total 62 30 (= 48%)

devant (in the front) 10 derrière (behind) 4 en face (across) 2 16 (= 25.8%)

Devant (in front of ) 10 Derrière (behind sth) 2 en face de (across from) 1 13 (= 21%)

derrière (behind) 3

POLISH za (behind) 15 total 71 przed (in front of ) 31

z tyłu (at the back) 2

w tyle (in the back) 1

Total 71

2 (= 2.8%)

za (behind) 16 przed (in front of ) 7 z tyłu (at back of ) 1 27 (= 38%)

37 (= 52.1%)

3 (= 4.8%)

1 (= 1.4%)

neutral expressions, i.e. prepositions like przed/za ‘in front of sth./behind sth.’ with the obligatory explicit mention of the relatum. Even if it is possible to produce an utterance like z przodu domu jest ulica ‘in the front of the house is a street’ where the explicit relatum is optional, native speakers of Polish describe this kind of scene with the preposition przed domem jest ulica ‘in front of the house is a street’. The obligatory explicit mention of the relatum-object linked to the use of prepositions involves reference maintenance in the domain of entities by the NP referring to the relatum-object. Thus, the concept of place is more important because the spatial span is delimited by this relatum-object and its outer boundaries. At the global level of reference (Table 9), native speakers of French more frequently employ object-specific expressions, such as au premier plan/au fond ‘in the foreground/in the background’, which encode the spatial span based on the intrinsic features of the relatum “poster”. Here, too, the concept of place is more important even though these expressions are often used intrasitively since the implicit relatum “poster” is clearly maintained constant. (14) au premier plan (0 = de l’affiche) il y a une rue principale in the foreground (0 = of the poster) there is a main street The relatum “poster” is not explicit but the “foreground” corresponds to the area of space linked to this specific relatum delimited by its outer boundaries. Native speakers of French rarely use object-neutral expressions like devant/derrière ‘in the front/behind’ in the global frame. Their use is more linked to the point-by-

 Marzena Watorek

point frame (cf. Table 10), where they are mostly used transitively with an explicit relatum. Here, the domain of entities is more important for anaphoric linkage. The comparison between the native speaker productions shows that in the global frame of reference, when the speaker refers to the global spatial span to localize the themes, the French speakers base their descriptions more on the placeoriented perspective where the referential domain of entities is essential. On the contrary, the Polish speakers base their discourse more on the space-oriented perspective. In the point-by-point frame, the speakers of the two languages prefer the place-oriented perspective and they accord more importance to the referential domain of entities in discourse cohesion. Let us now look at the learners’ data. Tables 11 and 12 show the repertoire of the locative expressions encoding saggital spatial relations in the global and pointby-point frames of reference for both groups of learners. Table 11. Repertoire of learners’ locative expressions involving the sagittal axis in the global frame of reference object-specific transitive intransitive

transitive

POL > FR (I) Total 2

assez loin (quite far) plus loin (farther away) 0 (= 0%)

0 (= 0%)

POL > FR (II)

Total 16

object-neutral intransitive

0 (= 0%)

sur le premier plan (in front) au premier plan (in the foreground) au fond (in the background) 0 (= 0%)

8 (= 50%)

2 (= 100%) plus loin (farther away)

0 (= 0%)

8 (= 50%)

Table 12. Repertoire of learners’ locative expressions involving the sagittal axis in the point-by-point frame of reference transitive

object-specific intransitive

object-neutral transitive intransitive

POL > FR (I)

devant (in front of ) 2

Total 2

2 (= 100%)

0 (= 0%)

0 (= 0%)

POL > FR (II)

devant (in front of ) 9

devant (in the front) 1

devant (in front of ) 2

Total 12

9 (= 75%)

1 (= 8.3%)

2 (= 16.7%)

0 (= 0%)

0 (= 0%)

The development of anaphoric means 

As for the learners in group I, we note a very limited number of spatial expressions encoding the spatial sagittal relations. In the global frame of reference two occurrences of object-neutral expressions used in their intransitive form assez/plus loin ‘quite far/farther’ refer to the global space without the intrinsic features of the relatum “poster”. This space is defined by the sagittal axis in terms of proximity in relation to the position of the speaker. In the point-by-point frame, two occurrences of the object-neutral expressions devant ‘in front of ’ are used in their transitive form and encode only the intrinsic projective relations where the relatum defines the spatial area. As mentioned above, the global frame of reference in the descriptions of the learners in group I is rather implicit (cf. Table 6) and their discourse is based on the point-by-point frame of reference. Projective relations are rarely expressed (cf. Table 1); the spatial configurations being described are conceptualised as topological relations, and their discourse is not very cohesive. We find in group I more than in group II a phenomenon of simplification of the information being transmitted in conjunction with the linguistic means that are most easily mobilised. These means, as we have seen, correspond to the simple syntactic pattern (B) V existential + NP where the locative expression – PP/ADV – is absent. The poor cohesion of the discourse can be reinforced by nonspecific means such as the additive particle aussi ‘also’ (cf. Table 8). These constructions are similar to those produced by intermediate Italian learners of French (Watorek 1996a), where the same syntactic pattern and the same interaction between this pattern and the informational structure of discourse were found. Hence, we note that these processes of simplification are shared by intermediate learners with two different source languages, Polish and Italian, acquiring the same target language, French. In the data of group II, object-specific expressions also occur in the global frame. However, though the same percentages of object-neutral and object-specific expressions are attested, we find a difference in the repertoire of the two types of expressions. In fact, the object-specific expressions are more diversified and their repertoire is similar to that of the native speakers (cf. Table 9). Despite an increase in frequency, the repertoire of object-neutral expressions is always the same: plus loin ‘farther away’. The other French object neutral expressions like devant/derrière ‘in front of/behind’ are never attested in the global frame. These expressions are only used transitively in the point-by-point frame. Expressions like devant/derrière seem to be considered by these learners as equivalent to the Polish prepositions za/przed used by Polish native speakers in the point-by-point frame. These learners do not perceive this equivalence between the French object-neutral expressions devant/derrière used by French native speakers in both the global and point-by-point frames, and the Polish object-neutral expressions z przodu/z tyłu used by the Polish native speakers only in the global frame. The devant/derrière (prepositions or

 Marzena Watorek

adverbs) in French have a larger semantic content that is distributed in Polish between the two pairs of expressions za/przed (prepositions only) and z tyłu/z przodu (prepositions or adverbs). Since za/przed are prepositions only, they are perceived as lexically equivalent to devant/derrière, which can either be prepositions or adverbs; devant/derrière are used by these learners only as prepositions. Thus, transfer of lexical equivalence goes hand in hand with transfer of the structural features of the expression.

.. Anaphoric linkage and the structure of the NP – relatum In both groups of learners’ productions, the domain of entities is dominant for reference maintenance. In expressions like devant where the NP referring to the relatum can be explicit or implicit, these learners prefer to make the relatum explicit. Thus, the NP in the spatial expression can be used for anaphoric linkage. An analysis of referential movement in learner and native speaker data shows the different ways in which the information is developed from one utterance to another. The following patterns can be distinguished: a.

Information is maintained immediately from the focus of one utterance to the topic of the next. The theme introduced in the first utterance is maintained as the relatum in the second. (15) 1 2

au milieu il y a [theme, focus une fontaine] in the middle there is [theme, focus a fountain] [relatum, topic à côté de la fontaine] il y a un arrêt de bus [relatum, topic next to the fountain] there is a bus stop

b. Information is maintained from the topic of one utterance to the topic of the following utterance. The relatum maintained in one utterance is maintained as relatum in the second one, as between utterances 2 and 3 in example (16): (16) 1 2 3

c.

au centre il y a une place in the middle there is a square [relatum, topic où] on voit une fontaine [relatum, topic where] we see a fountain [relatum, topic dans la place] il y a un tabac [relatum, topic in the square] there is a tabacco shop

Information is maintained immediately from the focus of one utterance to the topic of the other. The theme introduced in the first utterance is maintained in the following one as the theme as well. This is the case when the second

The development of anaphoric means 

utterance is the answer to the less frequent realisation of the quaestio, “Where is the X ?” (cf. Section 2). (17) 1 2

plus loin il y a [theme, focus un kiosque] farther away there is [theme, focus a newspaper stand] [theme, topic qui] est au centre de la place [theme, topic which] is in the center of the square

d. Information is re-introduced in an utterance at a distance. These are occurrences of types (a) or (b) without immediate maintenance. (18) 1 2 3

4

au milieu il y a [focus une place] in the middle there is [focus a square] où on voit une fontaine where we see a fountain à droite de la fontaine il y a un arrêt de bus avec des personnes qui attendent to the right of the fountain there is a bus stop with people who are waiting [topic dans cette place] il y a un kiosque à journaux et un marchand des fleurs [topic in this square] there is a newspaper stand and a flower seller

Our analysis of the French native data shows that the maintenance of the reference to entities is achieved by different NPs. The more distant an entity is in the discourse, the more complex the NP structure will be. We can indicate a continuum from the less presupposed entities to the more presupposed ones (cf. Figure 2 below). For re-introduction of an entity, NPs like Ce + N are used (cf. example (16)). Zero anaphore and the relative pronoun qui are used in utterances that answer the quaestio, “Where is the X ?”, and involve total presupposition (cf. example (17)). Actually, we have seen that these kinds of utterances augment the textual cohesion (cf. also example (8) and (9) above). A comparison between native speakers of Polish and French shows differences in the use of these different types of NP. One of the major differences between French: ce + N > le+N > personal pronoun (lui) > relative pronoun (qui) > 0-anaphore Polish: ten + N > N > personal pronoun (on) > relative pronoun (ktory) > 0-anaphore Figure 2.8

 Marzena Watorek

Table 13. Native speakers’ use of the demonstrative NP (ce/ten + N) for immediate maintenance Speakers

ce + N / ten + N immediate maintenance

ce + N other uses ten + N

FRENCH total 54

10 (= 18.5%)

44 (81.5%)

POLISH total 178

71 (= 39.8%)

107 (= 60.2%)

Table 14. Percentages of the different NPs used for the maintenance of reference to the entity-relatum

Polish and French lies in the specific grammatical means that mark the maintenance of reference in the domain of entities. French, but not Polish, has the definite articles le/la/les. Polish speakers optionally use the demonstrative articles ten/ta/to/ci. Thus, we find interesting differences in the use of the demonstratives ce/cette/ces in French and ten/ta/to/ci in Polish. The NP “ce + N” in French is essentially (81,5 %) reserved for the re-introduction of the referent (cf. ex. (18)). Only 18,5% of these NPs are used in the immediate maintenance of reference, which is usually marked by the definite article. On the contrary, in the Polish productions, 39,8% of these NPs (ten + N) are used with immediate maintenance (cf. Table 13). Returning to the learners’ data, Table 14 shows the use of the different NPs by the learners of both groups and, for comparison, by the native speakers. The use of the definite article for the expression of the maintenance of reference to entities increases from group I to group II. There is some evolution towards the target language (French). But the use of the demonstrative article equally increases in the productions of the learners in group II. Here, no evolution towards the target language can be attested. The tendency to resort to the cohesive organisation of the mother tongue is clearer, since demonstratives are also used more frequently in Polish native speaker production.

The development of anaphoric means

Table 15. The use of the NP “ce + N” for immediate maintenance of reference to the entitiy-relatum Speakers

ce + N immediate maintenance

ce + N other uses

POL > FR (I) total 11

2 (= 18.2%)

9 (= 81.8%)

POL > FR (II) total 40

11 (= 27.5%)

29 (= 72.5%)

When we look at the context in which demonstratives are used in the productions of both groups of learners, we note that their use is linked more clearly to immediate maintenance in group II than in group I (see Table 15). This developmental tendency differs from that found in the use of demonstratives in the French native speaker productions. In fact, the native speakers of French rarely used demonstratives for immediate maintenance of reference (see Table 13). A quantitative analysis leads one to believe that the learners of group I (intermediate) are closer to native French speakers than the learners of group II (advanced). However, a more detailed analysis of the contexts in which these forms are used reveals a development in anaphoric linkage that is independent of the phenomena identified in the two languages included in this study, Polish as the first language and French as the target language. The number of occurrences of demonstratives used by the learners of group I for immediate maintenance is extremely low (2 occurrences) because the anaphoric linkage in their descriptions is weak. The other uses of these forms do not correspond to the use of the native speakers of French, i.e. they do not correspond to the anaphoric use for the re-introduction of referents, but rather have a deictic value. These learners tend to use demonstratives to refer to entities that have not yet been introduced in the text but that are a part of the larger context. Of the 9 occurrences (other uses) 4 are used in this way as is illustrated in ex (19). (19) ANG, Polish > French (group I) 1 je vois une grande rue I see a big street 2 c’est une rue dans une ville This is a street in a town (. . . ) 3 il y a aussi un pont There is also a bridge 4 qui va d’un côté d’un bâtiment à une autre which goes from one side of a building to another



 Marzena Watorek

5 6

(. . . ) il y a aussi un monument (. . . ) there is also a monument c’est au centre de cette place it is in the centre of this square

In this part of the description, the entity square had not yet been introduced in the text. The learners of group I tend towards a non-anaphoric usage of demonstratives. Thus, their descriptions are based more on the extralinguistic context. In the descriptions of group II, the use of demonstratives increases. In general, they are used with a true anaphoric value. The use of demonstratives for immediate maintenance is greater than in the French native speakers’ productions and reflects the tendency of Polish native speakers. The native language of these learners has no article category. In their target language, French, the learners not only use definite articles to maintain reference; they also overgeneralize another form with anaphoric value from their native language, that is, the demonstrative adjective ten ‘this’. At first blush, we may interpret this as transfer from the L1. However, it is also possible to explain this phenomenon as a tendency to “over-mark” reference maintenance. This tendency is found in other studies of advanced learners. For example, Chini (1998) found that advanced German learners of Italian use “heavier” coreferential means like full NPs more than Italian native speakers. See also Hendriks (2003) who finds similar tendencies in L2 French and Gullberg (this volume). We observe that at first the learners of group I tend to presuppose referents excessively. Therefore the relatum is frequently omitted. The non-anaphoric use of demonstratives shows that these learners rely heavily on the extralinguistic context and little on their own discourse. Later, the learners of group II do not presuppose referents as much as they need to, and hence, we note an over-use of demonstratives for immediate maintenance. We find the same tendencies in studies on the development of cohesion in child discouse. Hickmann (1987) shows that children up to 7 years also presuppose the referent too much or do not presuppose it enough. These aspects of the child repertoire indicate that the referential system is mastered relatively late. The advanced learners of both groups, as adult speakers, can calculate mutual knowledge in referring, an ability developed during first language acquisition. However, the fact that these learners do not perfectly master the linguistic system of the target language influences their use of the referential system.

. Conclusions We have shown how learners at different levels of L2 acquisition construct a coherent and cohesive discourse, starting from a discourse where the anaphoric linkage

The development of anaphoric means 

STATIC MODE OF EXPRESSION Topological relations - AT PLACE ⇓ - INCLUSION ⇓ - NEIGHBOURING ⇓ - BOUNDARY

>

Projective relations -INTRINSIC ⇓ -DEICTIC

Figure 3. Spatial reference, acquisitional sequences, and complexification of spatial concepts expressed in spatial static descriptions by advanced learners

is relatively poor and progressing to a discourse where anaphoric linkage becomes more and more explicit. This kind of study points to an order for the expression of spatial concepts in discourse produced by intermediate and advanced learners. We have tried to show that this order is constrained by the evolution of the learners’ linguistic repertoire. Furthermore, we observe a parallel between the order in which the expression of the spatial concepts evolves in our study and the acquisitional sequences attested in the longitudinal study of second language acquisition by adult beginners in the ESF project (cf. Perdue 1993). Table 3 summarizes these parallel phenomena. Similar acquisitional sequences are also attested in studies on first language acquisition by children (cf. Johnston & Slobin 1979; Slobin 1993). We observe a similar development in child production where children first acquire the expression of the topological relations of inclusion and neighbouring, and then the expression of projective relations for entities that do not have an intrinsic orientation. The question is to know what motivates these shared paths. We can imagine that different factors and different reasons are at the origin of this convergence. For young children it is cognitive immaturity; they do not master projective relations because these relations are cognitively more complex. Adults learning an L2 have already acquired all sorts of spatial relations during first language acquisition and they master the cognitive complexity of these relations. The same acquisitional paths in the adults’ second language acquisition can be explained by the relation between the conceptually complex expressions and their use in learner – native speaker interactions. A topological expression relevant to the use of a theme and a relatum is in principle pragmatically less complex to use than an expression that encodes the projective relations where the speaker must make calculations based on his origo.

 Marzena Watorek

In the same way, the expression of projective intrinsic relations is less complex than the expression of projective deictic relations where the relatum is itself oriented.

Notes . Our data were collected at the Catholic University of Lublin in Poland with the help of Urszula Paprocka. . The poster used is part of the series “Hier fällt ein Haus, dort steht ein Kran und ewig droht der Baggerzahn, oder die Veränderung der Stadt” by Jörg Müller 1976, which was also used for data collection by Carroll and von Stutterheim (1997). . We use the terminology proposed by Klein (1985). The theme/relatum distiction corresponds to the figure/groud (cf. Talmy 1983) and to the cible/site (cf. Vandeloise 1986) distinctions. . The V – verb corresponds to the finite verb form. . The locative expressions include simple prepositions and complex prepositions (PP), adverbs (ADV) as well as lexical verbs and existential forms (il y a for French). . Entities like the poster or the picture are oriented on the sagittal axis. However, in the context of describing a poster, this axis is not pertinent because the elements behind the poster are not part of the spatial configuration being described. What corresponds to ‘in front’ coincides with ‘in front of the speaker’. . The expressions “w glab” ˛ and “w gl˛ebi” are not exactly synonimous. “W glab” ˛ is used more with the idea of movement as in “idac w glab ˛ (walking deep inside)” and “w gl˛ebi” is used more to express static location as in “on znajduje si˛e w gl˛ebi (he is deep inside)”. . These configurations seem to correspond to the universal of quantity proposed by Givon (1983).

References Bartning, I. (1997). L’apprenant dit avancé et son acquisition d’une langue étrangère. Tour d’horizon et esquisse d’une caractérisation de la variété avancée. AILE, 9, 9–50. Carroll, M. & Stutterheim, Ch. von (1997). Relations entre grammaticalisation et conceptualisation et implications sur l’acquisition d’une langue étrangère. AILE, 9, 83–116. Carroll, M., Murcia, J., Watorek, M., & Bendiscioli, S. (2000). The relevance of Information Organization to Second Language Acquisition Studies: The descriptive Discourse of Advanced Adult Learners of German. Studies in Second Language Acquisition, 22, 441– 466.

The development of anaphoric means 

Chini, M. (1999). Riferimento personale e strutturazione di testi narrativi in italofono e in apprendenti tedescofoni di italiano. In N. Dittmar & A. Giaccalone Ramat (Eds.), Grammatica e discorso. Studi sull’acquisizione dell’italiano e del tedesco (pp. 213–243). Tübingen: Stauffenburg. Givon, T. (1983). Topic continuity in discourse: An introduction. In T. Givon (Ed.), Topic continuity in discourse: A quantitative cross-language study (pp. 1–41). Amsterdam: Benjamins. Hendriks, H. (2003). Using nouns for reference maintenance: A seeming contradition in L2 discourse. In A. Giacalone Ramat (Ed.), Typology in Second Language Acquisition (pp. 292–326). Berlin, New York: Mouton de Gruyter. Hickmann, M. (1987). Ontogénèse de la cohésion dans le discours. In G. Piéraut-Le Bonniec (Ed.), Connaître et dire (pp. 239–262). Bruxelles: Ed. Pierre Mardaga. Johnston, J. R. & Slobin, D. (1979). The development of locative expressions in English, Italian, Serbo-croatian and Turkish. Journal of Child Language, 6, 529–545. Klein, W. (1985). Reference to space. A frame of analysis and some examples. Unpublished Manuscript, Nijmegen, Max-Planck-Institute for Psycholinguistics. Klein, W. & Stutterheim, Ch. von (1991). Text structure and referential movement. Sprache und Pragmatik, 22, 1–32. Perdue, C. (1993). Adult language acquisition: crosslinguistic perspectives. Cambridge: Cambridge University Press. Slobin, D. (1993). Adult language acquisition: a view from child language study. In C. Perdue (Ed.), Adult language acquisition, Vol II. Results. Cambridge Univ. Press. Talmy, L. (1983). How language structures space. In H. Pick & L. Acredolo (Eds.), Spatial orientations: theory, research, application (pp. 227–287). New York: Plenum Press. Vandeloise, C. (1986). L’Espace en français. Paris: Ed. du Seuil. Watorek, M. (1996a). Conceptualisation et representation linguistique de l’espace en italien et en français, langue maternelle et langue étrangère. Thèse en Linguistique, Université de Paris 8. Watorek, M. (1996b). Le traitement prototypique: définition et implications. Toegepaste Taalwetenschap in Artikelen, 55, 187–200. Watorek, M. (1998). L’expression de la localisation spatiale dans les productions des locuteurs natifs italophones et francophones, et d’apprenants italophones du français. Cahiers d’Acquisition et de pathologie du langage, 16(17), 17–50. Watorek, M. & Perdue, C. (1999). Additive particles and focus: observations from learner and native speaker production. Linguistics, 37(2), 297–323. Watorek, M. & Dimroth, Ch. (in press). Scope particles and Discourse. In H. Hendriks (Ed.), The Structure of Learner Varieties. Mouton de Gruyter.

Subject index

A agreement 16, 17, 20, 26, 33–35, 66 84–88, 97–99, 105, 106, 115, 119, 120, 122, 130, 151, 167, 292, 296 Aktionsart 56, 83, 171, 172, 174 anaphoric linkage 312, 315, 317, 320–325, 330–332, 340, 342, 343, 348, 351, 352 anchoring 2, 65, 66, 68, 89–91, 96, 115, 201, 291, 316 aspect 20, 49–57, 61, 85, 86, 102–105, 115, 120–122, 151, 172, 189, 191, 192, 201, 204, 207, 243, 245, 274, 275, 282 assertion 7, 16, 70, 75–80, 88, 89, 96, 99, 120, 133, 150, 159, 292 auxiliary 17–21, 24, 28, 34, 35, 45, 53–55, 58–60, 85–87, 90, 101, 121, 135, 136, 144, 149, 163, 169, 172, 175, 179, 203, 205, 240, 243,

B basic variety 19, 104, 106, 121, 126–130, 141–144, 165, 167, 190–192, 194, 197, 198, 206–208, 236, 238, 240, 242, 252–256

C case 17, 21, 26–28, 34, 35, 235, 292, 293 causal relations 279–281, 283 clitic pronoun 20, 21, 24, 26–28, 35, 240, 296 coherence 268, 270, 284, 332

cohesion 198, 291, 296, 301, 304, 311–313, 320, 322, 330–332, 340–343, 347, 349 communication strategy 322, 323 conceptualisation 109, 114, 116, 267, 293, 300, 301, 304, 305, 341 contrastive topic 151, 213, 225 copula 18, 19, 45, 50, 59, 60, 85, 131, 136–138, 142, 144, 159–180, 191, 338 cross-linguistic influence 222, 223, 228, 256, 276, 284 E ESF project 30, 32, 34, 37, 41, 51, 53, 70, 71, 103, 116, 119, 188, 240, 314, 319, 353 F finiteness 1, 2, 4, 15–21, 29, 31, 33, 36, 40, 65–67, 70, 72, 79, 84, 87, 88, 90, 91, 96, 105–107, 119–122, 129, 136, 137, 146, 149, 159, 160, 175–177, 179, 180, 232 focus 30, 51, 52, 69, 102, 104, 135, 150, 151, 213, 220, 230, 254, 269–271, 289, 292, 304, 305, 314–316, 319, 323, 330, 338, 348 fossilization 104, 116, 190, 205, 236 functional category 16, 21, 31, 36, 40, 41, 86, 87, 91, 120, 211, 230 G grammaticalisation 31, 106, 175, 192, 201, 208, 255

 Subject index

grammaticised meanings/functions 3, 49, 104, 106, 174, 182, 207, 267–270, 285 H habituality 83, 84, 105, 106, 111, 245 I imperfective aspect 49, 53, 54, 145, 170–172, 174, 189, 201–205, 245 infinite forms 20, 32, 45-47, 58, 85, 101, 108, 112–114, 150, 170, 171, 239, 243 254 INFL 16–19, 21, 23–25, 28, 31, 34–41, 91, 134, 215, 246 inflectional morphology 107, 113, 116, 177, 179, 235–238, 242 input 41, 46, 56–59, 62, 107, 110, 120, 126, 136, 179, 223, 251, 252, 255 iterativity 54, 83, 105, 106, 189, 194–199, 207, 208 L lexical verb 18, 19, 26, 45, 46, 48, 60, 66, 84, 87, 121, 128, 129, 137, 138, 144, 148–151, 173, 176, 191, 242, 243 linking 65–68, 70, 72, 76–84, 87, 101, 102, 105, 225 M macro-planning 10, 268, 283, 284 modal verb 20, 21, 25, 28, 34, 35, 59, 77, 8–83, 85, 86, 91, 113, 136, 141, 144, 149, 160, 218, 254 N negation 2, 16–19, 21, 23, 25, 26, 28, 30–36, 40, 72, 73, 78, 81, 87, 119–151, 163, 177–180, 245 non-finite verb 17, 21, 24, 33-39, 125, 137, 142, 148, 191, 192

P passive 246–248, 274, 275, 284, 290, 300, 302–304 perfect aspect 49, 53, 55–57, 87, 179, 192, 204, 205 perfective aspect 56, 131, 171–173, 177, 189, 204 post-basic variety 19, 121, 126, 129, 142–144, 150, 160, 165, 171, 178, 180, 190, 191, 196, 199, 203–207, 254, 255 pre-basic variety 30, 121, 142, 178, 190, 191, 194, 196, 106, 254 predication 50, 51, 53, 69, 70, 323, 324 presupposition 133, 134, 212, 349 progressive aspect 49–55, 174, 276, 281, 282

Q quaestio 98, 103, 104, 330, 337–339, 349

R reference introduction 51, 68, 109, 273, 279, 283 reference maintenance 4, 68, 269, 290, 291, 301, 312, 316, 317, 320, 321, 345, 348–352, 393 referential movement 311, 315, 324, 348 relational predicate 122, 148, 150 root infinitives 23, 24, 223

S scope particles 2–4, 77, 79–81, 89–91, 189, 194, 202, 208, 211–230, 336, 340 simplification 160, 163, 347 situation time 49, 83, 84, 96, 105, 179 subject 20, 21, 23–28, 32–36, 69, 105, 106, 214–216, 224, 225, 241,

Subject index 

244, 249, 254, 268–279, 282–284, 289–304 subordination 17, 21, 28, 36, 40, 274–277, 284 T temporal adverbial 83, 102–107, 110–113, 115, 116, 187–189, 191, 192, 197, 199, 204–207 temporal relation 99, 102, 104, 106, 115, 187, 188, 191, 192, 197, 206, 207, 242, 278, 281, 282 tense 3, 4, 17, 20, 28, 33, 39, 53, 54, 65, 66, 87, 88, 95–107, 109, 114, 115, 120–122, 129, 137, 151, 160, 163, 171, 172, 175, 180, 191, 200–205, 207, 242–251, 271, 274–276, 281, 282, 284 theme 298, 330, 331, 337–339, 348, 349 topic entity 68, 69, 96, 159, 290, 291–304

topic time 49, 54, 66, 83, 84, 88, 95–99, 103–107, 112–115, 150, 159, 175, 177, 179, 180, 207 transfer 18, 51, 57, 120, 126, 144, 178, 179, 222, 237, 348, 352 U time of utterance 99, 107, 159, 175, 179, 197 V variation 18, 37, 38, 61, 112, 150, 207, 211, 236, 240–244, 251, 254, 255 verb-second 59, 60, 84, 87, 88, 269, 282 Z zero anaphora 69, 291, 293, 296, 311, 320, 349

In the series STUDIES IN BILINGUALISM (SiBil) ISSN 0298-1533 the following titles have been published thus far or are scheduled for publication: 1. FASE, Willem, Koen JASPAERT and Sjaak KROON (eds): Maintenance and Loss of Minority Languages. 1992. 2. BOT, Kees de, Ralph B. GINSBERG and Claire KRAMSCH (eds): Foreign Language Research in Cross-Cultural Perspective. 1991. 3. DÖPKE, Susanne: One Parent - One Language. An interactional approach. 1992. 4. PAULSTON, Christina Bratt: Linguistic Minorities in Multilingual Settings. Implications for language policies.1994. 5. KLEIN, Wolfgang and Clive PERDUE: Utterance Structure. Developing grammars again. 6. SCHREUDER, Robert and Bert WELTENS (eds): The Bilingual Lexicon. 1993. 7. DIETRICH, Rainer, Wolfgang KLEIN and Colette NOYAU: The Acquisition of Temporality in a Second Language. 1995. 8. DAVIS, Kathryn Anne: Language Planning in Multilingual Contexts. Policies, communities, and schools in Luxembourg. Amsterdam/Philadelphia, 1994. 9. FREED, Barbara F. (ed.) Second Language Acquisition in a Study Abroad Context. 1995. 10. BAYLEY, Robert and Dennis R. PRESTON (eds): Second Language Acquisition and Linguistic Variation. 1996. 11. BECKER, Angelika and Mary CARROLL: The Acquisition of Spatial Relations in a Second Language. 1997. 12. HALMARI, Helena: Government and Codeswitching. Explaining American Finnish. 1997. 13. HOLLOWAY, Charles E.: Dialect Death. The case of Brule Spanish. 1997. 14. YOUNG, Richard and Agnes WEIYUN HE (eds): Talking and Testing. Discourse approaches to the assessment of oral proficiency. 1998. 15. PIENEMANN, Manfred: Language Processing and Second Language Development. Processability theory. 1998. 16. HUEBNER, Thom and Kathryn A. DAVIS (eds.): Sociopolitical Perspectives on Language Policy and Planning in the USA. 1999. 17. ELLIS, Rod: Learning a Second Language through Interaction. 1999. 18. PARADIS, Michel: Neurolinguistic Aspects of Bilingualism. n.y.p. 19. AMARA, Muhammad Hasan: Politics and Sociolinguistic Reflexes. Palestinian border villages. 1999. 20. POULISSE, Nanda: Slips of the Tongue. Speech errors in first and second language production. 1999 21. DÖPKE, Susanne (ed.): Cross-Linguistic Structures in Simultaneous Bilingualism. 2000 22. SALABERRY, M. Rafael: The Development of Past Tense Morphology in L2 Spanish. 2000. 23. VERHOEVEN, Ludo and Sven STRÖMQVIST (eds.): Narrative Development in a Multilingual Context. 2001. 24. SCHMID, Monika S.: First Language Attrition, Use and Maintenance. The case of German Jews in anglophone countries. 2002. 25. PILLER, Ingrid: Bilingual Couples Talk. The discursive construction of hybridity. 2002. 26. DIMROTH, Christine and Marianne STARREN (eds.): Information Structure and the Dynamics of Language Acquisition. 2003.