The Structure of Learner Varieties 9783110909593, 9783110184686

Citation preview

The Structure of Learner Varieties

W G DE

Studies on Language Acquisition 28

Editor Peter Jordens

Mouton de Gruyter Berlin · New York

The Structure of Learner Varieties

edited by Henriette Hendriks

Mouton de Gruyter Berlin · New York

M o u t o n de Gruyter (formerly M o u t o n , T h e Hague) is a Division of Walter de Gruyter G m b H & Co. KG, Berlin.

© Printed on acid-free paper which falls within the guidelines of the ANSI to ensure permanence and durability.

Library

of Congress

Cataloging-in-Vublication

Data

T h e structure of learner varieties / edited by Henriette Hendriks, p. cm. — (Studies on language acquisition ; 28) Includes bibliographical references and index. ISBN 3-1 1-018468-0 (cloth : alk. paper) 1. Language acquisition. 2. Language and languages — Variation. I. Hendriks, Henriette, 1962— II. Series. PI 18.SS14 2005 401Λ93 —dc22 2005011301

ISBN 3 11 018468 0 Bibliographic

information

published

by Die Deutsche

Bibliothek

Die Deutsche Bibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data is available in the Internet at < h t t p : / / d n b . d d b . d e > .

© Copyright 2005 by Walter de Gruyter G m b H & Co. KG, D-10785 Berlin. All rights reserved, including those of translation into foreign languages. N o p a r t of this book may be reproduced in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher. Cover design: Sigurd Wendland, Berlin. Printed in Germany.

Contents The ' Structure of Learner Varieties': Introduction to the volume Henriette

Hendriks

Part I: Referential movement Reference to persons and objects in the function of subject in Learner Varieties Bernt A hrenholz

19

Reference to person in learner discourse Marina Chini Structuring space in discourse: A comparison of Chinese, English, French and German LI and English, French and German L2 acquisition Henriette

111

Hendriks

Two dimensions of the representation of complex events structures: granularity and condensation. Towards a typology of textual production in LI and L2 Colette Noyau, Christina de Lorenzo, Maria Kihlstedt, Urszula Paprocka, Gema Sanz Espinar, and Ricarda Schneider Cross-linguistic analysis of temporal perspectives in text production Christiane von Stutterheim and Monique

Lambert

Determinants in first and second language acquisition: person, space, and time in discourse across languages Maya

Hickmann

231

vi

Contents

Part II: Scope The semantic knowledge base for the acquisition of negation

263

and the acquisition of finiteness Angelika

Becker

The acquisition of negation in Italian L2 Giuliano

315

Bernini

The acquisition of negation in French L2. An analysis of

355

Moroccan Arabic and Spanish "learner varieties" Patrizia Giuliano and Daniel

Veronique

Additive and Restrictive Particles in Italian as a Second

405

Language. Embedding in the verbal utterance structure Cecilia

Andorno

Additive Scope Particles in Advanced Learner and Native

461

Speaker Discourse Marzena

Watorek and Christine

Dimroth

Reading from outside: Acquisitional patterns in a cross-

489

linguistic approach Davide Index

Ricca 505

The Structure of Learner Varieties: Introduction to the volume1 Henriette Hendriks 1. A way of looking at (second) language acquisition Over the last few decades, the field of second language research has grown dramatically. Theoretical approaches are plenty and are often linked to very different interests. Some were born out of an interest in the relation between language and society and how this affects acquisition, some try and look for universal vs. language-specific influences on acquisition, some try to explain acquisition from a functional perspective, a cognitive perspective, and yet others try to the availability of a language acquisition device (in the UG sense) to second language learners. This introductory chapter serves to make explicit the framework in which the contributors to this volume operate, to introduce some common terminology and definitions, and to give the reader a sneak preview of the articles to follow. In 1997, Klein and Perdue advocated the following perspective for language acquisition research, as characterised by four assumptions: 1. During the acquisitional process, the learner passes through a series of learner varieties. Both the internal organization of each variety at a given time as well as the transition from one variety to the next are essentially systematic in nature. 2. There is a limited set of organizational principles of different kinds which are present in all learner varieties. The actual structure of an utterance in a learner variety is determined by a particular interaction of these principles. This kind of interaction may vary, depending on various factors, such as the learner's source language. With successive input analysis, the interaction changes over time [...]. 3. Under this perspective, learner varieties are not imperfect imitations of a 'real' language, the target language, but systems in their own right, errorfree by definition, and characterised by a particular lexical repertoire and by a particular interaction of organizational principles. Fully developed languages are simply borderline cases of learner varieties. They represent a relatively stable state of language acquisition - that state where learners stop learning because there is no difference between their variety and the input, the variety of their social environment.

2

Henriette Hendriks 4. If all learner varieties, including the final one, are manifestations of the human language capacity, then the study of this capacity should not start with the most complex of these manifestations, and go from there to the simpler ones. Rather, it is advisable first to study the various organizational principles of human language and their interplay in relatively simple cases, those where the various form-function relations are more elementary and transparent.

Some of these assumptions have been around in second language acquisition research for a long time. Corder (1967) first advocated this type of approach, and Selinker (1972), who baptised learner varieties with the name interlcingnage, gave the approach its more common name. Some of the assumptions, however, and in particular the latter part of the third assumption and the fourth are more specific to Klein and Perdue. This volume presents ten articles studying and discussing the structure of such learner varieties at various levels of the proficiency scale (very early learner varieties, the Basic Variety, and varieties which are so close to the borderline / target language variety that we call speakers of those varieties nearnatives or quasi-bilinguals) and in a range of source-target language combinations (from Polish/German to Moroccan-Arabic/French). The learners are mostly non-guided learners, that is, they learn the language through everyday communication, in a natural fashion, free from systematic and / or intentional guidance. The authors of the articles all adhere to the learner variety approach, as it is called and thus analyse the structure of such varieties as systems in their own right, error-free by definition. They search for the logic underlying those systems as well as for the logic underlying the transitions from one system (or variety) to the next one. The articles have originated from a five-year research project, a follow-up to the ESF financed project on Second Language Acquisition by Adult Immigrants (Perdue 1993) 2 . The project was known under the name "The Structure of Learner Varieties", hence the title of this volume, and all contributors (the authors of the discussion articles excepted) participated in it. Apart from the above four assumptions, one more common assumption reunites and directs the research in this group. It is felt that the main function of language is communication, and that given this fact, language acquisition should preferably be studied in situations in which the learner is trying to use language for communicative purposes (rather than studying the learner's capacity to fill in gaps in an exercise, or to judge the acceptability of certain target language constructions for example). As Klein (1986) already argued, the non-guided learner is in a seemingly paradoxical situation as far as

The Structure of Learner Varieties: Introduction

3

communication is concerned. In order to communicate, he will have to learn some of the language, but for language learning this particular type of learner only has communication as a reliable tool. Of course, Klein points out, this is only seemingly the case, because human beings have many more tools at their disposition for communication than just language (facial expression, gestures, etc.). Still, in many cases language is clearly the most effective and precise tool. Producing language is a complex activity. If we take a look at Levelt's (1989) model of language production, for example, it is suggested that there are four components that all have to be mastered at least to some extent in order for a verbal message (read communication) to be successful: conceptualisation; formulation; articulation and self-monitoring. During the conceptualisation phase, preverbal messages are being generated. It is during this phase that the speaker has to take decisions concerning the intent of the message, the selection of information allowing that intent to be realised, and the linearisation of the information at utterance and text level. These decisions taken, this leads to the existence of a so-called preverbal message. The preverbal message serves as input for the formulation phase. During this phase, items are extracted from the mental lexicon and grammatical relations reflecting the conceptual message are being generated. The product of this set of operations is called the surface structure, which again serves as input for the phonological encoding. During the articulation phase, the phonetic plan is realised in form of a series of instructions for the articulatory organs, thus resulting in external discourse. Finally, during the entire production process the speaker monitors the correspondence between his communicative intentions and his internal and external discourse. This allows him to detect any eventual correspondence problems, and to provide selfcorrections. Given the complexity of the entire production process, Levelt (1982) proposes to use so-called complex verbal tasks as experimental material for the study of language in use. Complex verbal tasks imply the production of extended, coherent discourse. Such tasks (film retellings, route-directions, stage instructions) allow the researcher to observe all four components of language production, given that a speaker will have to make choices concerning the information to be provided for the communicative intent and the interlocutor involved; to linearise the information according to discourse-pragmatic principles; and to encode all such information with the linguistic means as available in the target language. Native speakers have been shown to occasionally have problems with any of the four components of the production task, so it is to be expected that learners will have such problems as well.

4

Henriätte

Hendriks

All research in this volume has used complex verbal tasks to elicit its data across languages. The project as a whole is therefore discourse oriented more than sentence oriented in its approach. The articles, within a common framework to be discussed below, all address the following general questions, which show the functionalist interest of the researchers, questions asked in the hope to get a better insight in the organisational principles of learner varieties and the constraints on those principles: 1. How does the learner express and integrate information from different semantic domains (time, space, persons, events), when producing a coherent text at a given time? 2. How do his procedures change over time? 3. Which causal factors (cognitive, age-related, universal vs. language specific) determine these changes?

2. Previous project: what we can take as "given" As mentioned above, the Structure of Learner Varieties project was preceded by a large cross-linguistic project financed by the ESF and coordinated at the Max-Planck Institute for Psycholinguistics in Nijmegen. This project took the issue in the fourth assumption above seriously, i.e., that in order to understand more about the human language capacity one should start by looking at very early varieties rather than looking at fully-fledged varieties. It analysed the very first and early acquisition of non-guided second language learners, just after their arrival in the country of the target language with no prior knowledge of that target language (cf. Perdue 1993 for a detailed discussion of the project, the subjects and results). The main results of the previous project can be summarised as follows. First, a limited set of organisational principles operating in the learner varieties could be identified. These were mainly constraints on three levels: phrasal, semantic and pragmatic. The constraints were found to interact, and it is this interaction that determines the actual organisation of a learner variety at a given point in time. The kind of interaction, and hence the specific contribution of each principle may vary depending on source language influences or on the level of proficiency of the learner. As a result, the interaction may change over time. Second, learners seemed to pass through three stages. Overall they started out with the so-called nominal utterance organisation (also pre-basic variety, Perdue 1996). At this stage, spontaneous utterances would mainly involve seemingly unconnected nouns, adverbs and particles. The main reason for the

The Structure of Learner Varieties: Introduction

5

impression of disconnectedness is probably the lack of verbs, which would have allowed to impose some kind of implicit organisation on the elements in the utterance, for example through shared knowledge about the argument structure of the verb used. This stage was followed by an infinite utterance organisation, or the so-called Basic Variety. In the Basic Variety, recognizable verbs start to occur in utterances. As mentioned, this presence of verbs allows the learner to implicitly or explicitly assign roles in the argument structure. The Basic Variety was found to be at least partly independent of any of the languages under study. It represents a self-standing if limited system of communication, which has a universal status (at least for the ten linguistic cases of acquisition studied). As such, it provides insight into a 'basic' procedure for organising information in (connected) speech available to any adult speaker, learners or natives (for more elaborate descriptions of the basic variety, cf. Klein and Perdue (1992, 1997)). As long as the semantic, pragmatic and phrasal constraints that organise it can be reconciled, the Basic Variety works. However, when the sets of constraints contradict each other this provides an excellent motive for learners to acquire more specific structuring devices for their target languages, and they move towards a finite utterance organisation (post-basic variety). This third phase can be characterised by the occurrence of the distinction between the finite and non-finite components of the verb (hence it's name). Not all learners in the project reached this stage, given that the Basic Variety turned out to be a point at which learners seem to fossilise more easily The transition from one stage to the next was always found to be slow and gradual. Finally, it seemed to be the case that initial steps in development were dominantly guided by universal principles, whereas factors attributable to specifics of source and target languages seemed more characteristic of later stages of development.

3. The new project: The Structure of Learner Varieties The new project and hence work presented in the present volume, built on the findings available from the first project. It now had a reliable way of characterizing learner varieties at different proficiency levels, knowing what possible utterance structures are available at these various stages, what pragmatic and semantic constraints rest on a learner's production at a given stage, and what the meaning of utterances may be when structurally constraint in a particular way. The Structure of Learner Varieties (SLV) project also made

6

Henriette Hendriks

extensive use of the data base that resulted from the previous project (as will become clear in the volume). The organisation of the SLV project, however, was very different from its predecessor. Where the ESF project had been very strictly organised in terms of data collection, type of analyses, etc., the Structure of Learner Varieties project was more loosely organised. Not every one made use of the same database; not all members researched learner varieties of similar proficiency levels, etc. The project set out to further the research in two specific sub-domains: Referential movement in texts, and the use of scope in learner varieties. It wanted to understand more about the interaction of principles organizing the macrostructure of the text, and those principles structuring individual utterances. The sub-domain of scope was chosen because it had not been dealt with at all in the ESF project, and it does involve this particular link between macro- and micro structure of text (see below for more detail). Referential movement had been looked at before, and was furthered in this project by looking more specifically at the integration of information from different conceptual domains (time, space, person, event, and modality) in the speaker's construction of a coherent text, by looking at a larger range of learner varieties, comparing adult second language learners with child monolingual and bilingual learners (thereby introducing the age factor), and by looking in more depth at issues that were found to be important during the ESF project, but for which no time had been left to study them more thoroughly (for example the issue of granularity).

4. Part One: Referential movement in language production The first part of the volume on referential movement includes five articles, and is concluded by an article discussing the issues from the five articles from a point of view outside of second language acquisition (Hickmann). Where the topics dealt with in this part of the volume may seem to be rather diverse: reference to person (Ahrenholz; Chini), reference to space (Hendriks), and reference to time/events (Noyau et al.; von Stutterheim and Lambert), all of them are dealt with within one common framework, known as the Quaestio model. As a result, articles know a common terminology / language, and comparability of the analyses and results is high and straightforward.

The Structure of Learner Varieties: Introduction

4.1. The Quaestio

1

model

The Quaestio model as formulated by Klein and von Stutterheim (1987) provides an interface between the conceptualisation and the formulation of a text (cf. also Levelt's production model as discussed above), and accounts for both coherence and cohesion of texts. It is an analytic framework for the crosslinguistic studies of languages, and was used as a basis of analysis also during the ESF-project. According to the Quaestio model, the structure of all coherent texts is constrained on both global and local levels by the nature of the (implicit or explicit) question - the Quaestio - which the text in its entirety is meant to answer. Examples of such questions are: "What happened to Charlie Chaplin"?, "What entities are where on the poster"?, etc. According to this approach, any utterance in the text integrates information from a combination of domains (person, space, time, events, and modality), the particular Quaestio influencing the way in which possible domains of reference are realised in utterance and text and the development of the domains across utterances. As a result, a narrative will, for example, be mainly organised through the temporal domain, three other domains also being crucially involved, however, namely: spaces, persons (and objects), and events. For a route description, the domain of space will more important. The Quaestio thus imposes constraints upon (a) the macro structure of the text, which concerns aspects such as linearisation of information, and the partitioning of utterances into background and foreground. But it also imposes constraints on (b) the concrete form of the individual utterance, where it largely determines which information in the utterance is new and which information is maintained; moreover, it (largely) determines which part of the utterance is topic information and which part is focus information. This two-fold partitioning of the entire information to be expressed has strong consequences on linguistic form. To exemplify this, consider the following communicative situation: When asked (the Quaestio) "What did you do on holiday this year?" a speaker could answer: "We went to Scheveningen on the Dutch coast and spent our time swimming, walking along the beach, and eating fish. We had a very nice apartment". This answer contains various bits of information, some of which seem to directly answer the Quaestio, i.e., we went to the Dutch coast, we swam, we walked, we ate fish, and others which do not seem to answer the Quaestio, but rather give some other relevant information (we had a nice apartment). Those parts of the response directly answering the Quaestio (we swam, ate fish, etc.) are qualified as foreground (the skeleton of the text) by the framework, whereas the other parts are qualified as background.

8

Henriette

Hendriks

The Quaestio also influences how information is introduced and maintained in a text, in that objects, for example, will usually be introduced and spatially contextualised in the foreground in living room descriptions, whereas they may occur in the background in another type of text. The Quaestio defines topic and focus within each utterance in that, as any question, it raises a set of alternatives. The answer will specify one of these alternatives. The set of possible alternatives raised by the Quaestio is called the topic, and the one alternative which is selected and specified is the focus of the utterance. Thus, to return to the above example, the topic component in the answer will include the protagonists and the time given (we and reference to the past), whereas all the events listed (go, swim, eat, etc.) form the focus.

4.2. The individual articles in short The focus in this first part of the volume will be on the basic variety and development beyond. In earlier projects on referential movement, a lot of attention has been given to the acquisition of the individual linguistic means necessary to refer, and to their very early uses in discourse whilst the learner tries to execute the complex communicative tasks he is confronted with. Those (earlier) projects made detailed analyses, amongst others, of the various linguistic means needed to refer to persons and entities, to spatial situations and to sequence events in time, and they have provided us with the knowledge about acquisition orders of the encoding of spatial concepts, the gradual buildup of the noun phrase, knowledge about the order of acquisition and use of pronouns, and order of acquisition of verbal morphology and its usage by learners for marking subject-verb agreement and temporal information. With proficiency, the two types of knowledge to be acquired 1. language specific forms and grammaticalisation 2. matching of the various functions on utterance and discourse level with these language-specific forms change in order of importance. At the lower proficiency levels the acquisition of forms is immanent. However, the more proficient the learner becomes, the more one can take his knowledge of the target language forms for granted. Advanced learners do not, as von Stutterheim and Lambert phrase it (this volume), "have to struggle for words, but master the formal system to a larger extent". At that point, matching forms with functions in a target-like manner (taking into account the pluri-functionality of the forms) becomes the main job

The Structure of Learner Varieties: Introduction

9

of the language learner. One of the specific hypotheses to be tested in the light of this idea is the presupposition that, having acquired the grammatical means of the target language, a further level of proficiency may be identified for those learners who use the means for creating discourse cohesion following the target language patterns. As a consequence, the data in this part, although not excluding early varieties, concentrates mainly on learners who are attempting to or have advanced beyond the basic variety, an attempt seen in this project as motivated by two reasons: a need to escape from the expressive constraints of a simpler system on the one hand, and a wish to match the target-linguistic environment more closely on the other hand. Furthermore, all papers focus more on the acquisition of the target-language specific discourse functions held by the particular forms acquired by the learners, rather than by the acquisition of the forms as such. Thus, the papers by Ahrenholz and Chini not only discuss the acquisition of forms for referring to person and objects as acquired by the learners, but, more specifically, discuss how source and target languages differ in their choice of particular forms for discourse cohesion, and test hypotheses concerning the particular problem the learners may have to appropriately match form and function in the TL as a function of markedness and saliency of the forms, typological differences between source and target languages (± prodrop), etc. Hendriks' paper on reference to space is concerned with discourse cohesion as reinforced by the presentation of spatial information in narratives, rather than with the acquisition of the actual forms. Again, it focusses on choices of forms at the utterance level which are motivated either by language specific differences or by a more global discourse level organisation (cf. also von Stutterheim and Lambert), i.e., decisions concerning the level of explicitness of the spatial locations referred to, and the explicitness of the packaging of spatial information. The main questions asked are: how well and how fast do child learners of a first vs. adult learners of a second language adjust to target language specific patterns in this domain of reference where the underlying discourse pragmatic principles are similar, but their rendering in language is variable. This paper thus introduces another variable, namely, the effect of age of the learner. Finally, the papers by Noyau et al., and von Stutterheim and Lambert, although both mainly focussing on reference to time and events in discourse, in fact provide us with studies of the organisation of information structure on a more global (reference domain independent) level. Both investigate the conceptualisation and verbalisation of complex event representations in texts, given a constant, non-linguistic complex of events. Thus, Noyau et al. propose to look at two key aspects of the conceptualisation of event structures in texts, namely granularity (the degree of temporal

10

Henriette

Hendriks

partitioning of a situation), and condensation (the degree of hierarchical organisation of event structures). These authors hypothesise that the degree of granularity is closely related to the development of lexical means and will thus allow to distinguish early proficiency levels of language acquisition. Once the acquisition of linguistic forms stabilised, granularity will become a free variable for the more advanced learners as it is for native speakers of a language. Condensation will be important in more advanced learner varieties only, and it is hypothesised that learners although they will show a general development from accumulative presentation of information towards a gradual command of hypotaxis, will be influenced by LI usage of condensation rather than follow the patterns of the L2. By comparing different learners (child vs. adult L2) Noyau et al., similar to Hendriks, introduce the age variable. Finally, von Stutterheim and Lambert are interested in the decision taken by very advanced learners (a variety still largely unknown) as to what events to select for verbalisation and how to present the events once chosen. They list a number of options that are available in various languages for the representation of events, and look at the native and non-native choices of these options, the options being: - topic-time management (how are situations related to time) - level of granularity - choice of the amount of information specified per event unit - perspectivation of the events. They conclude that whereas "these learners clearly master the syntactic constraints holding between lexical items at sentence level, the way they select and present information and relate events is evocative of global strategies of LI discourse organisation". The discussion paper by Hickmann, who is looking at the five articles from an acquisitional (but not just second language acquisitional) perspective, discusses how the articles contribute to ongoing research, focusing on the following three central questions in acquisition research; 1. What are universal vs. language-specific aspects of acquisition? 2. What are language-internal structural vs. functional determinants of acquisition? And 3. what are language-external determinants of acquisition?

The Structure of Learner Varieties: Introduction

11

5. Part Two: Scope in learner varieties Studies concerning scope are being dealt with in the second part of the volume. It includes five papers and a discussion paper which puts the findings in a perspective exterior to second language acquisition (Ricca). Scope phenomena traditionally include: -

negation additive and restrictive scope particles such as only, also, and even frequency adverbials such as often, mostly, twice, several times, and quantified expressions such as many books, all girls.

Some of these phenomena (especially scope particles) have been rather neglected in second language acquisition research. To some extent, this may be a consequence of the fact that detailed knowledge about the interpretation of such scope phenomena is lacking even in the target languages, which in turn may be a result of the fact that scope phenomena concern optional rather than obligatory elements of utterances, that is, although these elements fulfil important communicative functions, they are structurally not "necessary". As a result, they will occur less frequently in languages (less true for negation), and may therefore have been discussed less frequently. They present an important learning problem, though, in that L2 learners have to reconcile the logical structure of their utterances with the syntactic specifics of the language being learned. In the following articles, therefore, two of the above phenomena will be studied in more detail: negation in Italian as an L2 (Bernini), French as an L2 (Giuliano and Veronique), and German as an L2 (Becker), and additive and restrictive scope particles in Italian as an L2 (Andorno) and in German and French advanced L2 varieties (Watorek and Dimroth). The Structure of Learner Varieties, as a project, set out to answer the following questions: 1. Are there circumlocutory possibilities for expressing the meaning carried by a scope particle before that particle is acquired? 2. What is the order of acquisition of scope particles? Does one observe a similar order of acquisition across languages, and if so, why? 3. Is there a developmental order in the use of utterance-level particles versus constituent-level particles, and in observed combinations of particles? 4. What are the principles governing the use of negation and scope particles in context at a given moment of the acquisition process? Is it the case, for example, that the learner first uses an iconic placing of the particle adjacent

12

Henriette Hendriks

to the material under scope, before the syntactic specifics of the relevant target language are acquired, i.e., is there a basic use of negation and scope particles showing strong cross-linguistic similarities? The articles concerning scopal features all set out to look at one or more of the above questions and other more specific questions. In order to find answers, researchers in the project all worked, again, within one framework, as proposed by Dimroth and Klein (1996). The framework is based on a number of premises, the main one of which being that scopal elements, despite them being optional, follow clear structural constraints.

5.1.

The

framework

According to Dimroth and Klein, five concepts are necessary in order to adequately describe scope phenomena in a given language. The first of those concepts is the so-called preliminary structure (Ausgangsstruktur). This is the pre-existing structure on which the optional scope element operates and which meaning it modifies to a certain extent. The second concept is the potential or maximal scope (Wirkungsbereich). This is the part of the preliminary structure that can be effected by the optional element. This potential or maximum scope, which depends on the position of the scope particle and, in some languages (e.g., German) on its stress, is considered to be a syntactic phenomenon in the target language. The integration of the scopal element into the preliminary structure and the thereby defined domain of application (scope) may therefore be more or less straightforward, depending on the language in question. Languages, moreover, allow a further restriction of the elements within the scope of the particle that are actually affected by the particle. When this happens, this more restricted group of elements, called the focus (Fokus), is a part of the particle's maximal scope, and can be especially accentuated through language specific means such as intonation or word-order. It is important to stress that the delimitation of this actual domain of application depends on the information structure of the relevant utterance in context. Particles can only affect information that is contrastive with respect to earlier utterances, that is, information that meets one of the classical definitions of focus. An assumption made by Dimroth and Klein (1996) is that the information structure of a preliminary structure doesn't change through the insertion of a particle. It is thus the preliminary structure's focal or contrastive part that can be affected by a particle. Independent of this, the particle's maximal scope can include other elements.

The Structure of Learner Varieties: Introduction

13

A further concept in the Dimroth and Klein framework is the so-called set of alternatives (Auswahlmenge) (cf. also the Quaestio model above). The set of alternatives consists of all meaningful elements that can sensibly replace the Fokus part of the preliminary structure. Note however, that, even when theoretically sensible, not all elements will always be in this set. The set will usually be restricted by the utterance context and world knowledge. In the preliminary structure, the speaker choses one of the members of the set of alternatives. The scopal particle placed in the preliminary structure informs us of the relation between that chosen element and all other elements in the set of alternatives, that is, are there other elements in the set for which the utterance can be valid, are there more likely elements in the set than the one in the utterance, is the element in the utterance the only one possible, etc. Which of the relations we are concerned with is expressed in the lexical meaning of the particles. To give an example: (1)

a. John drank a beer b. John only drank a beer c. John also drank a beer

We can consider la to be a preliminary structure. That being the case, lb shows an added optional scope particle, only. It is inserted before the finite verb. The domain of application lies to the right and is adjacent to the scope particle. It thus includes "drink a beer". Included in the set of alternatives may be all actions for which John is the agent: play the piano, eat with chopsticks, etc. The lexical meaning of the scope particle only expresses that only the chosen element of the set of alternatives corresponds to what John did. None of the other possible elements of the set would accurately describe John's actions. In lc, all structural features stay the same, but the lexical meaning of the particle also suggests that there are other activities in the set of alternatives that John has been the agent of, apart from drinking a beer. Both for the description of scope phenomena in a given language, and in order to acquire a certain language, two problems have to be solved: a. One has to discover the lexical meaning of the particular scope particle. b. One has to discover how that particular particle interacts with the preliminary structure i.e., which position/stress pattern corresponds to which scope. When studying fully-fledged varieties (target languages), the possible preliminary structures existing for the speaker to choose from, including Fokus

14

Henriette Hendriks

phenomena, even though sometimes extremely complex, in principle are known to the linguist. In learner varieties, it will be of essence for the linguist to get himself acquainted first and foremost with the scala of preliminary structures available at a given point in time to the learner, the semantic and pragmatic sense expressed by such structures, and the existence of Fokusstructures available to the particular learner. Only after having established this knowledge can one start speculating about such a learner's use of scope elements. The ESF project, having studied precisely this type of information for the learner varieties involved (cf. section 2 above), has provided the researchers with a reliable way of characterizing the learner varieties along those line. All the contributors to this volume use the categorisation proposed in the earlier project into pre-basic variety learners, basic variety and post-basic variety learners. Meaning and structural qualities of the individual particles cannot be presupposed known to either LI or L2 learners, nor, often, to the linguist attempting to describe a full grown target language. Therefore Dimroth and Klein propose to not try and code occurrences using a pre-fabricated coding scheme, but rather to list possible meanings and uses. They suggest that, when accumulating evidence in such a way, patterns will naturally become clear to the researcher. The advantage of proceeding in this way is that one is not likely to be guided by preconstructed beliefs about the functioning of scope particles either in the specific target languages, or in languages in general.

5.2. The individual papers in short Given that all five articles have used this framework, this introduction should make reading of them easier. As you will find by reading the articles in detail, mostly questions 2 to 4 were answered. Overall results show that both negation and additive and restrictive particles occur very early in the learner data, irrespective of the fact that they are optional. This shows how important their communicative functions must be. The project being of a cross-linguistic kind, you will also find that the results of the various articles allow one to recognise universal vs. language-specific trends very clearly. As far as negation is concerned, all three articles (Bernini, Giuliano and Veronique and Becker) allow the conclusion that learners, irrespective of the source and target languages will start out using holophrastic negation. Again, across languages, some initial basic universal principles seem to reign at the early stages of acquisition, making the learner place the negator adjacent and to

The Structure of Learner Varieties: Introduction

15

the left of the negated element. This strategy has been claimed to be consistent with cognitive principles, and it is also the cross-linguistically most frequent placing of the negator. Depending on the actual placement of the negator in the target language (pre-finite in Italian, post-finite in French and German), the acquisitional path is obviously going to differ, in that not much structural development is found in Italian (Bernini), but development is clearly found in French and German, where it is suggested that development might be tied to the development of fhiiteness on the one hand (Giuliano and Veronique; Becker), and be influenced by the source language on the other (Giuliano and Veronique). Findings concerning the acquisition of scope particles in Italian (Andorno) show that here too the basic universal and maybe cognitively guided principle "operator immediately in front of the affected element" is valid. The need to maintain the group consisting of the particle and its affected element is stronger than other principles which seem to prevail in languages such as French, i.e., "do not interrupt the basic structure". The principle does occur early in acquisition when scope particles are placed at the external positions of the utterance, but later on in acquisition the group principle prevails. The work by Watorek and Dimroth, finally, takes the questions a step further in that they do not look at the interaction between the scope particle and the preliminary structure on the utterance level, but they go beyond that level, and try to analyse the function of scope particles on the discourse level, thereby linking the first and second parts of this volume. Learners in this study are relatively far advanced (using Bartning's 1991 criteria), in that their language production in many respects is similar to language produced by native speakers, except that the learners show areas of fragility in the domain of discourse. It has been shown previously (Watorek and Perdue, 1999) that additive particles can play a role in discourse cohesion in that they allow speakers to implicitly maintain reference to a same entity, as exemplified in (2). The square, although not explicitly mentioned again, can be easily inferred as the meant location for the trees. (2)

In the square is a fountain And there are also some trees

The usage of additive particles in learner languages for such purposes forms the subject of the Watorek and Dimroth study. Results show that if such simplifying solutions as in (2) above which permit the omission of additional spatial information to be made explicit, are readily available in the target language (for example in French L2), learners will prefer such options over the

16

Henriette

Hendriks

more explicit (and therefore more complex) ones. However, when a simplification exists but is highly infrequent and constraint by other syntactic rules, learners will chose not to use such structures even though it forces them to acquire more complex utterance structures (German L2). Finally, w e invited Davide Ricca to discuss this second set of articles, Ricca bringing in a more overall linguistic rather than acquisitional perspective. Ricca discusses various questions in the area of scope, such as the role of typological markedness in acquisition, and the role of intonational saliency in the acquisition of sentence negation. He also compares the acquisitional paths as found in replacing negation and focus particles, thereby providing an explicit comparison of the t w o scopal elements.

Notes 1. I would like to thank Christine Dimroth for her valuable input concerning the description of the framework used in the scope project. Thanks also go to Tim Hoy for checking my English. All remaining mistakes are, of course, mine. 2. Note that the project is also well known as the "ESF project", and may be referred to as such in all the articles in this book, and even outside of it by non project members.

References Bartning, Inge 1991 L'apprenant dit avance et son acquisition d'une langue etrangere. Tour d'horizon et esquisse d'une caracterisation de la variete avancee. Acquisition et Interaction en Langue Etrangere 9: 9-50. Corder, Pit 1967 Significance of learners' errors. International Review of Applied Linguistics 5: 162-169. Dimroth, Christine and Wolfgang Klein 1996 Fokuspartikeln in Lernervarietäten. Ein Analyserahmen und einige Beispiele. Zeitschrift für Literaturwissenschaft und Linguistik 104: 73-114. Klein, Wolfgang 1986 Second Language Acquisition. Cambridge: Cambridge University Press. Klein, Wolfgang and Clive Perdue 1992 Utterance Structure: Developing Grammars Again. Amsterdam: John Benjamins.

The Structure of Learner Varieties: Introduction

17

Klein, Wolfgang and Clive Perdue 1997 The Basic Variety or: Couldn't natural languages be much simpler? Second Language Research 13,4: 301-347. Klein, Wolfgang and Christiane von Stutterheim 1987 Quaestio un referentielle Bewegung in Erzählungen. Linguistische Berichte 109: 163-183. Levelt, Willem 1982 Linearization in describing spatial networks. In: Stanley Peters and Esa Saarinen (eds.), Processes, Beliefs and Questions, 199-220. Dordrecht: Reidel. Levelt, Willem 1989 Speaking: From Intention to Articulation. Cambridge, Massachusetts: MIT Press. Perdue, Clive (ed.) 1993 Second Language Acquisition by Adult Immigrants (2 vol.). Cambridge: Cambridge University Press. Perdue, Clive 1996 Pre-basic varieties: the first stages of second language acquisition. Toegepaste Taalwetenschap in Artikelen 55: 135-150. Selinker, Larry 1972 Interlanguage. In: International Review of Applied Linguistics 10:209-231. Watorek, Marzena and Clive Perdue 1999 Additive particles and focus. Observations from learner and native speaker production. Linguistics 37,2: 297-329.

Part I: Referential movement

Reference to persons and objects in the function of subject in Learner Varieties Bernt Ahrenholz

1. Introduction How do language learners refer to persons and objects in the function of subject? More specifically, which pronominal means do speakers of pro-drop languages use when they acquire a non-pro-drop language such as German? The present article reports the findings of a case study of an Italian learner of German. The theoretical framework is provided by the quaestio model and its categories of "reference", "referential movement", "topic" and "focus". The Quaestio model distinguishes five domains of reference, one of which is reference to persons and objects in the function of subject. 1 In German there are a variety of ways to refer to this domain: names, nouns, nominal phrases and various kinds of pronouns can function as explicit references. Implicit reference, expressed by zero anaphora and corresponding morphological marking of the verb, is only possible under very restricted conditions in German, whereas it occurs frequently in pro-drop languages. It is therefore of particular interest to analyse how learners of pro-drop languages (like Italian) acquire languages in which explicit reference to the subject is obligatory (like German). In the present paper, the following aspects will be explored in detail: -

which linguistic devices can be used for the domain of reference "persons and objects in the function of subject" in German and Italian, which pronominal devices are used by an Italian learner of German, how these devices are acquired over time, which functions do these devices have.

Reasons for the observed developments in learner varieties will also be discussed.

20

Bernt Ahrenholz

2. The domain of reference "Persons and objects in the function of subject": Linguistic devices in German and Italian Firstly, various possible ways of realising the subject function in German will be described in as far as they are relevant to the analysis of the learner varieties under investigation. 2 For this analysis, it is useful to distinguish between ways of referring to the first and second person on the one hand and the third person on the other. To a certain extent, this distinction reflects the distinction between deictic and anaphoric pronouns, which has proven to be important in studies on second language acquisition (see below). 3

2.1. First and second 2.1.1. First person

person

In German, the pronouns ich in the singular and wir in the plural are obligatory for the first person. 4 However, there are some exceptions: 5 -

(1)

-

-

When the referent is unambiguous, in the prefield, 6 subject ellipsis of the first and second person singular may occur under certain interactional conditions: kommst du0 [ich] komme gleich7 come you? - 0 [I] come immediately 'Are you coming? [I'll] be right there' There are a number of formulaic expressions with no explicit subject reference, e.g., forthe first person singular: verstehe '[I] understand', gratuliere 'congratulations'. Subject omission also occurs in co-ordinated sentences under certain conditions (see example 2 below and section 1.3. for more detail).

2.1.2. Second person In general, the second person must also be referred to explicitly. 8 The respective pronouns are du 'you' in the singular, ihr 'you' in the plural, and the polite Sie 'you' in the singular and plural. Again, there are some exceptions: - Subject ellipsis is acceptable in certain co-ordinated sentences.

Reference to the subject in German L2 (2)

-

jetzt now und and 'Now

21

nimmst du die aktentasche take you the briefcase 0 [du] gehst zu dem grünschwarzen stuhl 0 [you] go to the green-black chair you take the briefcase and go to the green-black chair'

With the exception of Sie, there is no explicit lexical reference to the second person (singular/plural) in imperatives:

(3)

second person singular: second person plural:

komm 0 'come kommt 0 'come

bittel please!' bittel please!'

2.1.3. Italian The Italian pronominal system is very similar to the corresponding German system; even the sound of some of the pronouns is similar (ich - io, du - tu). A central difference, however, is that the use of pronouns is rarely obligatory in Italian. Consequently, free pronouns are often not realised. 9 (4)

Vieni ? - Vengo! come? - come! 'Are you coming? - 1 am coming!'

Grammatical descriptions of the Italian pronominal system thus refer mainly to contexts in which these pronouns are realised. According to Cordin and Calabrese (1988), the Italian pronoun has a number of different functions, including the following: 10 -

it it it it

is used to emphasise the referent, can be necessary with verb ellipsis, occurs when the utterance is co-ordinated with another nominal phrase, occurs as expression of surprise.

Thus Italian personal pronouns are used especially for emphasis, disambiguation of referents, topic-focus structuring, and other such syntactic purposes: 11

22

Bernt Ahrenholz

(5)NS: 1 2

EXP: NS:

per

spedire il pacchetto + 0 dovresti to send the parcel (you) should tirarlo fnori e costruirlo (h) pull it out (of the bag) and construct it mhm e io dovrei dirti come si fa a costruire and I have to tell you how to construct (it) (NS Giusi, instruction "constructing a parcel")

2.2. Third person 2.2.1. German There are a number of possible ways of referring to the third person in German: 13 -

noun phrases (names, nouns, more or less complex noun phrases and numerals), the anaphoric use of personal pronouns (er, sie, es for the three genders in the singular and sie for all genders in the plural), the deictic use of personal pronouns, the pronominal anaphoric use of der, die or das for the three genders in the singular and die for all genders in the plural, the deictic use of der, die, or das, the pronominal use of the demonstrative pronoun (dieser, diese, dieses), the pronominal use of the indefinite article {einer, eine, eins),14

Further possibilities are: -

the expletive es, the impersonal pronoun man, the reflexive pronoun sich, zero anaphora.

The most important contexts for zero anaphora in the third person are again co-ordinated clauses. As the rules operating here are not very easy for learners to understand, the conditions for sentence co-ordination with subject ellipsis will be explained in greater detail in section 1.3.

Reference to the subject in German L2

23

2.2.2. Italian In Italian, it is again the case that third person pronouns are generally not used if it is obvious to whom the utterance refers. 15 This applies to maintained reference as well as to reintroduced or shifted information. (6)NS:

questo:: frate (h)voleva eh:: vuole eh:: this monk wanted wants rapire il il bambino to kidnap the the child EXP: mh(hm) NS: quindiche 0 e entrato + eh:: in casa therefore [he] is entered into (the) house dalla finestra rompendo il vetro through (the) window breaking the glass (NS Stefano, narrative 'doll's house')

Similar to the first and second person, third person pronouns are mainly used for emphasis, to disambiguate reference, or to underline the topic-focus structure. 16

2.3. Zero anaphora in German In German, the omission of explicit lexical reference to the subject is limited to the context of sentence co-ordination and topic continuity, as shown in the following example: (7)NS:

jetzt nimmst du die aktentasche (re-introduced) now take you the briefcase und 0 gehst zu dem grünschwarzen stuhl (maint.) and 0 go to the green-black chair und 0 stellst die aktentasche links neben dich and 0 put the briefcase left beside you (maint.) 'Now you take the briefcase / and (you) go to the green-black chair / and (you) put the briefcase on your left' (NS Christa, instruction 'ashtray')

There are some constraints on constructions such as this with co-ordination and zero anaphora: 17

24 -

-

-

Bernt Ahrenholz At the beginning of the construction there must be an overt subject, that is, the subject must have the status of "newly introduced" or "re-introduced" information. In example (7) the pronoun du has this status. The subject must be part of the topic 1 8 (this is always the case in instructions such as example (7) if the subject refers to the agent of the indicated activity). The subject must have the status of "maintained" information in all subsequent utterances, that is, the subject must remain constant. The lexical marking of the subject can only be dropped in utterances which are dependent on the same quaestio and in general follow on directly. The position before the verb cannot be occupied, because it is already given to a subject which is not phonetically realised. 19

Such constructions with zero anaphora can be very long, as in the following example taken from a retelling of a film with Charlie Chaplin 2 0 : (8)NS:

und

kantine in eine (maintained, start of 'canteen' segment) and he goes to the canteen in a 0 nimmt sich ein grosses tablett (maint.) 0 takes himself a big tray 0 lädt sich ungefähr zwanzig teller darauf (maint.) 0 puts himself about twenty plates on it 0 packt die alle voll mit essen (maint.) 0 piles up them all full of food 0 setzt sich hin (maint.) 0 sits himself down 0 speist gemütlich (maint.) 0 eats unhurriedly (maint.) theke und 0 seht dann vorne an die counter and 0 goes then front to the 0 kann nicht bezahlen (maint.) pay can not 0 ruft noch einen polizisten von draussen herein (maint.) __ calls yet a policeman from outside in und 0 sast ihm (maint.) and 0 tells him

0

er

seht

an

die

Reference to the subject in German L2 also

dass er die

rechnimg

Well that he the bill und 0 wird dann abgeführt

25

nicht begleichen kann (maint., side structure, subord.) not pay can

(maint., end of'canteen' segment) and 0 is then taken away (NS Christa, film retelling Charlie Chaplin) There are different reasons for the use of constructions like the one above: - They create high text cohesion. - They allow the speaker to segment different parts of ongoing events and bundle them into one complex event. However, zero anaphora are not as optional as one might think. Explicit reference to the subject in these constructions would lead to a slight shift in meaning: (9) (7')

jetzt nimmst du die aktentasche (re-introduced) now take you the briefcase und du gehst zu dem grünschwarzen stuhl (maint.) and you go to the green-black chair und du stellst die aktentasche links neben dich (maint.) and you put the briefcase left next you 'Now you take the briefcase and you go to the green-black chair and you put the briefcase on the left beside you'

If the utterance were produced as in example (9), the subject would become part of the focus, even if it is not stressed. The meaning would be to underline that this specific subject, rather than anyone else, has to fulfil the instructions - a possibility which is excluded in the contextual setting.21 Moreover, there are some restrictions on the omission of the subject in German; the most important are as follows: -

If the connector und is used with an adverbial (e.g., und jetzt 'and now', und dann 'and then', und so 'and therefore'), inversion occurs and the subject has to be realised explicitly:

26

(10)

Bernt Ahrenholz

NS:

er rennt noch hinter-m zag herA he runs still after-the train aber 0 kriegt ihn nich mehr but 0 gets him not more und dann versucht er halt And then tries he auf anderen wegen dahin zu kommen on different ways there to come 'He runs after the train, but doesn't make it, and then he tries to get there in another way' (NS Paul, film retelling Harald Lloyd)

However, it is possible to use adverbs in a different position, i.e., after the verb. This permits subject-drop constructions. This phenomenon is illustrated in example (8) (und 0 geht dann vorne an die theke). - In secondary sentences, the explicit lexical use of the subject is always required, whether or not it is part of the main structure: (11) NS:

nächste szene is next scene is wie charlie dann auf der strasse steht (re-introd.) how Charlie then on the street stands und 0 nich wees (maint.) and (he) not know was er machen soll (maint., side structure) what he do should 'The next scene is how Charlie is then standing on the street and (he) does not know what he should do' (NS Martin, film retelling Charlie Chaplin)

3. Previous research Non-target-language-like subject omission was reported as a specific feature of some German learner varieties already in very early studies (cf. Clyne 1968; HPD 1977; Keim 1984), but only few studies have centred on the acquisition of reference to the subject in German (Klein and Rieck 1982). Nevertheless, there are some studies, especially investigations of syntax, in which this aspect is included in the analysis. Klein and Rieck (1982) studied the acquisition of personal pronouns by Italian and Spanish migrants on the basis of data

Reference to the subject in German L2

27

collected from 27 informants in the Heidelberg corpus. Rieck (1989) studied nine adult Spanish learners of German in a longitudinal study. In the ZISA project, the acquisition of the subject reference was investigated in conjunction with the development of syntactic structures (Clahsen, Meisel, and Pienemann 1983). A study conducted by Pienemann (1981 and 1998) on the acquisition of German by three eight-year-old Italian girls took a similar approach. Kuhberg (1990) studied two eleven-year-olds with Turkish and Polish as their source languages. Broeder's extensive study of the acquisition of pronouns by Turkish and Moroccan learners of Dutch (Broeder 1991) also considered the acquisition of German possessive pronouns by Turkish and Italian learners. And finally, Ulbricht (1987) discussed the conditions under which subjects can be omitted, based on an error analysis of the written and oral productions of Hungarian learners of German, who often tend to omit the subject reference in subordinate clauses. Finally, Kuhs (1989) studied the written German narratives composed by 18 children of Greek immigrants. Nearly all of these studies considered the acquisition of German by learners with a pro-drop source language. It seems that there are barely any studies of learners with a non-pro-drop source language such as Dutch or English. We were only able to find a single case study of an Australian student learning German in Pienemann (1998). 22 Diehl et al. (2000) have studied Frenchspeaking pupils. 23 The most detailed study on the acquisition of German pronouns was conducted by Klein and Rieck (1982). This research on the Heidelberg corpus revealed that: -

deictic pronouns are acquired before anaphoric ones, the article das is the first anaphoric form used, anaphoric reference is partly substituted by the anaphoric use of the nouns, not al 1 pronouns are restricted to their function in the target language (ich is also used for the first person plural), there is a slow change from omission of the reference to explicit reference, with both forms coexisting for a long time, there are learner-specific functions of pronouns which do not occur in the target language (mir for subject reference), during stages of acquisition in which learners dispose of appropriate devices for explicit reference, cases of target-language subject omissions, used frequently by the native speakers, are rare in the data.

28

Bernt Ahrenholz

The late development of anaphoric devices is probably connected to the fact that the pronouns in question are not very salient; they are not necessarily perceived at all in the input. Based on the findings of Klein and Rieck (1982), Klein (1990) argued against the UG view, stating that the obvious difference in the acquisition of deictic and anaphoric pronouns suggests that functional aspects are more important for the acquisition of second languages than syntactic ones. In her discussion of Klein (1990), Lalleman (1993) argued that deictic and anaphoric pronouns have different functions; deictic pronouns identify a person, and there is no other choice of linguistic device, whereas anaphoric pronouns refer to persons, objects, etc., and there is a choice of possible devices. In her "functional approach to markedness", pronouns are seen as being more marked than nouns, which explains why they are acquired later. Dittmar and Skiba's (1992) investigation of the acquisition of German by three Polish learners (P-MoLL project, cf. Dittmar, Reich, Schumacher, Skiba and Terborg 1990) also found that ich and du were used in the very first recordings, whereas the third person pronouns were not used until much later. 24 Pienemann (1979) and Clahsen, Meisel, and Pienemann (1983) explored the extent to which the omission of the subject is linked to the development of syntactic structures. The temporary omission of the subject was seen as an attempt to avoid syntactic conflicts involving a change in word order ("permutation") before inversion has been acquired. Other cases of subject omission were interpreted as "reduction of the grammatical system" (Clahsen, Meisel, and Pienemann 1983: 196; similarly, Pienemann 1998). Kuhs' (1989) attempt to corroborate these findings with her data on Greek learners failed because there were only very few subject omissions, and only three of them could be interpreted in the context of word order. 25 The findings of Breeder (1991), who conducted a study into the acquisition of Dutch within the ESF Adult Language Acquisition project, were similar to those of Klein and Rieck (1982): the pronouns for the first person singular are acquired first, followed by those for the second, and then those for the third person singular. Singular forms are used before plurals (this applies to the first/second person and, with certain restrictions, to the third person), masculine forms tend to be used before feminine forms, and subject forms before object forms. 26 Furthermore, reduced forms seemed to be used quite late in learner varieties, and were often produced as unanalysed chunks (cf. Young-Scholten 1993). Hong (1995; cf. also Clahsen and Hong 1995) investigated the problem of zero anaphora. On the basis of grammaticality judgements and reaction time

Reference to the subject in German L2

29

tests of Korean learners of German, H o n g attempted to show that, contrary to the assumptions of parameter theory, there is no relation between subject-verbagreement and the acquisition of zero anaphora. Davies (1996) was also unable to demonstrate a relationship between the use of zero anaphora and subject-verb-agreement in his research on English as a second language. Hendriks (1998) analysed reference to persons and spatial reference in the narratives of Chinese learners of German (and, for comparative purposes, the acquisition of German and Chinese as LI by children), and showed that Chinese learners of German use mainly syntactic devices to differentiate between new and given information. Klein and Perdue (1992) conducted a comprehensive study on the development of syntactic structures in early learner varieties. Their research involved Italian and Turkish learners of German, and focused on the functional, semantic and syntactic principles of the production of utterances. Where the use of pronominal devices is concerned, marked differences between learners can be observed. Some learners acquire the complete pronominal system, whereas others have barely any anaphoric devices at their disposal.

4. The data The case study to be presented here is based on data provided by an Italian learner of German (part of the longitudinal P - M o L L research project conducted by Norbert Dittmar; cf. Dittmar et al. 1990). 27 The data for the Italian adult learner Franca include 21 recordings made over a period of VA years. Oral productions were recorded for the discourse types "conversation, narrative, instruction, statement, problem solving, and description of pictures", each with a series of different tasks. The discourse types and tasks were implemented in a three-cycle rotational design, making it possible to compare data f r o m an early, a mid and a late stage of the acquisition process. Although the tasks were not designed for the elicitation of pronouns exclusively, the large amount of them ensures obligatory contexts for most of the pronouns to occur at least every second month. The only pronouns for which these obligatory contexts are not necessarily frequently present are the 2 nd person plural Ihr and the politeness form Sie. For the present study, the whole corpus was analysed with respect to pronouns in the function of subject using the computer program WordCruncher (cf. Müller 1992). Furthermore, the utterance units of the first cycle (about 3000) as well as the narratives recorded in the 26 th and 35 th month were cate-

30

Bernt A hrenholz

gorised with respect to the subject function in order to ensure that nonpronominal devices and subject omissions were also captured.

5. Results 5.1. Deictical use of pronomina Similar to learners in the HPD (cf. Klein and Rieck 1982) and P-MoLL projects (Skiba and Dittmar 1992) and learners of Dutch in the ESF project (Breeder 1991), the Franca data reveal a striking difference in the acquisition of deictic and anaphoric pronouns: Deictic pronouns are acquired at a very early stage, anaphoric pronouns distinctly later (cf. Tables 1 and 2 in the appendix). In Franca's variety the pronouns ich Τ and du 'you' as well as wir 'we' appear markedly before er 'he' and sie 'she, they'. With respect to the pronouns which can be used anaphorically or deictically, we find only a late deictic usage of the article das in the Franca data, whereas the pronominal usage of the articles der and die is restricted to anaphoric functions.

5.1.1. ich The pronoun ich is used from the very beginning of the recordings. The context (free conversation and Franca's narration of her biography as an immigrant) provides ample opportunity for the frequent use of ich in the very first recording (72 occurrences; see Table l). 28 Indeed, ich is the most frequently used pronoun in the whole corpus (1536 occurrences). As shown above, neither the source language nor the target language provide options other than the explicit (sometimes clitic) realisation of the pronoun or the omission of the subject. (12)NNS:

ehm ich will-eh mh+ gehen in-eh schuleΛ Elim I want-eh mh+ go in-eh school ehm *in ottobre %novembre*% in october/ november wenn-eh+ 0 [ich] kommzuriickA ehm aus *italia* when come back from Italy Ί want to go to a school in October or November when (I) come back from Italy' (NNS Franca, 4th month, free conversation) 29

Reference to the subject in German L2

31

As shown in this example, ich is always used to denote the first person singular. In addition to the frequent use of ich, however, subject omission of the first person singular occurs occasionally until the 18th month (cf. examples 12, 13 and Figure 1 in appendix). From the perspective of the target language, this is to be seen as a learner-specific use of subject omission. 30 Overall, the proportion of ich omissions is so small in the first months that certain explanations of this phenomenon, such as the connection with syntactic rules assumed in the ZISA project, do not seem to apply here. 31 When the referential movement is taken into account, however, it is striking that in all cases of ich omission, the reference is either maintained (example 12) or reintroduced. Hence, it does not seem to be a pure transfer phenomenon, as subject omission is also possible in Italian when a referent is introduced for the first time. Franca also omits ich in later stages of acquisition, but only in formulaic expressions such as weiss nich 'don't know' which also occur in the spoken German of native speakers (cf. Auer 1993). 32 (13)NNS:

eh ja eh diese: kleine eh schwarze *coperchio* yes this small black lid @@ _ [ich] weiss nich η: name [I] know not name eh kannsdu eh 1/ eh lassenhier can you leave here 'Yes, this small black lid - 1 don't know what it's called - you can leave it here' (NNS Franca, 14th month, "coffee machine" instruction)

5.1.2. da The pronoun du for the second person singular is found from the very beginning of the data (cf. Table 1 and Figure 2 in appendix) and, from the semantic point of view, it is always employed as in the target language. The pronoun is used 487 times in the whole corpus, most often as subject pronoun in connection with a verb. 33 Very often, it is used in instructional discourses in which an experimenter is told how to perform a specific action (see below or example 15). In the other discourse types, du is used mainly in questions — even in the first recording:

32

Bernt A hrenholz

(14)NNS:

?kennst du l know you ? 'Do you know ?' (NNS Franca, 4th month, free conversation)

Although du is used from the very first recordings (4th month, example 14) it should be noted that the pronoun is used only rarely up to the 11th month. This may be partly due to the fact that utterances with du tend to be side structures 34 in most discourse types, and therefore are not to be expected in high numbers. But the same cannot be said of the instructions performed in the 6th, 11th and following months. Here, a limited number of subject references with du are to be expected on account of the tasks given; most of the instructions are agent-oriented (Ahrenholz 1998a; Kohlmann 1997; von Stutterheim 1997), i.e., most instructions (given by both native speakers and learners) refer to the listener as the person carrying out the action35, as shown in the following example: (15) NNS:

du nehmen die tasche you take the briefcase du setzen oben stuhl you put on top chair 'You take the briefcase you put it up chair' (NNS Franca, 6th month, instruction "ashtray", 1st cycle)

Indeed, most of the du occurrences in the Franca corpus are to be found here. Accordingly, the recordings with no instructions (21st, 24th and 34th months in Figure 2) have a very low rate of du occurrences. The use of du is not only dependent on the type of discourse, however. This is shown above, in the first instruction of the 6th month. Here, there are only few verbs in the lexicon, and verb morphology has not yet developed. By the 11th month, a richer verb lexicon and the acquisition of verb morphology seem to fulfil the preconditions for a broader use of du.

5.1.3. wir, ihr and Sie The pronoun wir 'we' for the first person plural also occurs in Franca's variety from the very beginning. With a total of 100 occurrences in the Franca corpus, this pronoun is primarily used to refer to the joint activities of the informant and her partner. Furthermore, a total of 15 occurrences of the lexeme ihr 'you'

Reference to the subject in German L2

33

(plural) can be observed from the 4 th month on; it is mainly used as possessive pronoun, and is only once used as subject noun in the 31 st month. 36 Because the informal and friendly du was used in the contact between informants and experimenters, the more formal and polite pronoun Sie 'you' (singular and plural) does not appear in the data with the exception of two role plays recorded in the 31 st and 36 th month. On these occasions, 38 occurrences of Sie were generated. As the pronouns were realised without hesitation, selfcorrection or misuse, they can be regarded as already acquired. 5.1.4. der, die and das Where der, die and das 'the, this' are concerned, the initial and most frequent use is in the function of the definite article (in 1612 of 1979 occurrences; cf. Ahrenholz 2004 for a detailed discussion). They also occur as relative pronouns (in 82 occurrences) and as pronouns (283 occurrences) which are used anaphorically and deictically (cf. Tables 1 and 2). But only das seems to have the same sort of relevance as a pronoun that it does as an article (464 occurrences, 247 as a pronoun). Almost from the outset, das is used in "presentationals" to identify objects or persons in descriptions of pictures and in instructions. (16)NNS:

eh:ff+3+

das ist-e eine bar that is a bar 'That's a bar' (NNS Franca, 5th month, description of picture)

Das is also used in questions, and later on it is used deictically in the context of instructions. 37

5.2. Anaphorically

used

pronouns

In the Franca data, the pronouns er, sie (singular and plural), es, der, die and das are used anaphorically. 38 All anaphoric pronouns are acquired distinctly later than the deictic ones. The most important alternative in this period seems to be the use of non-target-language-like subject omission; the use of noun phrases, observed in other studies (cf. Klein and Rieck 1982 or Diehl et al. 2000) in the context of maintained reference, is not very frequent (cf. Figure 6 in appendix). From the 14th month on, the pronouns sie, er and das (cf. Table 2 and Figure 6) are used particularly often, whereas there is a marked drop in the number of NPs and non-target-language-like subject omissions.

34

Bernt Ahrenholz

5.2.1 er There is scattered usage of the pronoun er as early as the 4th and 5th month, but it is used more frequently from the 20th month on. It is striking that the pronoun er is used so rarely in the first cycle, because the narratives "Doll's House" (8th month) and "Charlie Chaplin" (14th month) offer ample opportunities for its use, as shown by the relatively frequent occurrences in the second (24th and 26th month) and third (34th and 35th month) cycle (cf. Figure 3 in appendix). Accordingly, there are numerous contexts in the first months of acquisition in which an explicit lexical reference by means of a noun or pronoun is required, but is missing (cf. Figure 3): +2+ ntein bruderA ehm + oh ah jetz-e ist-eh my brother now is *non so come si di/ consulente finanziarioA* not (I) know how say consultant financial EXP: mhm NNS: ehm + arbeitA work eh 0 ist *sposato* mit eine kinder fünf jähre [he] is *married* with one children five years 'My brother is now working as a - 1 don't know how you say it financial consultant (and he) is married with one child, aged five' (NNS Franca, 4th month, free conversation)

(17)NNS:

Here, as in almost all cases, examination of the referential movement shows that the reference is "maintained" or "reintroduced". In other cases, subject omissions correspond to the use of subject omissions with co-ordinated utterances and referential identities in the target language. The difficulties in the acquisition of anaphoric pronouns such as er are not only apparent in their scarcity of use and frequent omission. In the first recordings, various indications of learner problems are perceptible: for example, in the 4th month, Franca uses er with repetition as well as the corresponding Italian pronoun lui: (18) NNS:

und-eh ehm + e/+ and eh hab/ habe. mh + have

erA + %mhA erA *lui* mhA%er he he *lui* he gespra/sprachen %gespra% (h) spoken

Reference to the subject in German L2

35

mit-ehm + 3 + mit-e _ with with @ [I] forgot @mit ehm + lente eh *di with people * o f * 'And he has spoken with the people from ' (NNS Franca, 4th month, free conversation) In another case, there is double reference with both a name and a personal pronoun. Finally, there are occasional self-corrections of the alternatives er and sie:39 (19) NNS:

und so eh fur ehm + eine arbeit-eh and so for a job wie eh finden (h) se/ als Sekretärin like as find secretary er/ e./ sie eh + muss + spreche/ gut sprechenA he she must speak well speak 'And so she must speak well to find a job as secretary' (NNS Franca, 18th month, free conversation)

As in this example, all self-corrections concerning the choice of the pronoun are successful. With respect to the function, it can thus be stated that, in all cases, references made to previously introduced persons or objects correspond to the chosen pronoun in terms of gender. 40

5.2.2. sie The lexeme sie has various functions in German. It is used as a deictic pronoun in the singular and plural in polite forms of address (see above). For feminine nouns, it is used to refer to persons (and objects) in the singular, to either persons or objects in the function of subject or accusative object. In the plural, sie is used to refer to nouns of all genders in the nominative and accusative cases. In Franca's variety, sie is first used as a pronoun for the third person plural in the 9 th month. This is the case in the following example, where Franca is telling the experimenter that her brother and his family have been to visit her: (20) NNS:

oh aber but

eh + jetz now

ein a

bisschen eh ehm bit

besser better

36

Bernt A hrenholz + eh eh fur-e ehm + weihnacht (h) e/ + for Christmas sie alle komm/ ehm + hatte gekommen (h)ier they all come had come here 'Now it's a little bit better / they all came here for Christmas' (NNS Franca, 9 th month, free conversation)

The pronoun sie is frequently used for the plural from the 11th month on, and for the singular from the 14th month on. This sequence of occurrence in the data may be incidental. On the other hand, it may be linked to the inherent tendency for the plural form to be used more frequently, as sie can refer to all three genders in the plural. Then again, the fact that the singular form occurs later could also indicate difficulties with respect to the inflexion of the verb. Whereas the plural form corresponds to the unmarked form ending in "-en ", the third person singular ends with an "-t", and sometimes requires a change of vowel in the stem: (21) NNS:

EXP: NNS:

die madchenA-eh the girl und soA ehm_ +4+ and so ?stohltA %stohlt stoles stoles

ist-eh auch-eh ungrig_(h) is also hungry ge/ eh:m sie eh g/ hm +1+ eh she + gestohlt &stohlt% ?& stolen stoles

&ja stiehlt& Yes steals *stielt A-eh ein-ehehm ein brot in einen wagen + steals a bread in a car 'The girl is also hungry, so she steals a loaf of bread from out of a car' (NNS Franca, 14th month, Charlie Chaplin retelling)

Even when the pronoun sie can be considered acquired, it is not used in all obligatory contexts (cf. Figure 4 in appendix): (22) NNS:

und-eh: charlie chaplinA-eh mit-eh and charlie chaplin with mit der mädchenA (h) ehm + fa/eh with the girl

die mädchen the girl fallen-eh: fall

Reference to the subject in German L2

37

auch raus-%eh% in diestrasseA (h) also out into thestreet und soA können-eh 0 weg-egehen + in: + frei and so can away go in free 'And Charlie Chaplin and the girl fall out into the street, and then (they) can get away' (NNS Franca, 14th month, Charlie Chaplin retelling) The subject omission of sie is not only observed in plural uses (such as the example above), but also occurs in singular cases. Apart from these nontarget-language-like subject omissions, there are some cases of subject omission in connection with co-ordinated clauses which would also be possible in the target language: (23) NNS:

aber e/m/ wann-eh eine person + @*bestemmia*@ but when a person *curse* oder (It)eh 0 is hose'' or is angry 'But when a person curses, or is angry' (NNS Franca, 9 th month, free conversation)

In nearly all cases, sie refers to females or to objects with feminine gender. However, there are a few occurrences in which sie is used to refer to an institution, i.e., an impersonal agent. In these cases, although it would be possible to use sie in the sense of the third person plural in German, German native speakers would use the pronoun die (cf. Ahrenholz 2004) or a passive construction instead. Example (24), which deals with the issuing of a new passport at a consulate, illustrates this use of sie in the Franca variety: (24) NNS:

un-eh ichwarte dass neuer papier neuer and I wait that new paper new passport ehm machen eh sie machen für mich passport make eh they make for me 'And I'm waiting for them to prepare my new papers, my new passport' (NNS Franca, 11th month, problem solving)

38

Bernt Ahrenholz

5.2.3. es Es has the following functions: a) an anaphoric pronominal reference to nominal phrases with the neuter gender, b) an expletive es in sentences such as es ist gut 'it's good', c) a formal placeholder if the subject is part of the focus (es can be omitted under certain conditions here), d) a formal placeholder for subject or object phrases. Sometimes c) and d) are also considered "expletive". 41 Without entering into a discussion on questions of terminology, in the present analysis a distinction is made between pronominal anaphorical uses on the one hand, and expletive (including syntactical) uses on the other. In the first case es has a clear semantic function, whereas in the second case it is semantically empty. For the acquisition of German as a second language it may also be of importance to know that in Italian there is no neuter gender, that es in German is not very salient phonetically, and that it is often used clitically in spoken language. The Franca corpus comprises 205 occurrences of es, including 19 anaphoric pronouns, 7 placeholders in the prefield, 124 expletives (primarily in es gibt) and 55 clitic occurrences. Only in late stages of acquisition is the pronoun used more frequently. Up to the 18th month, es is used on only 7 of the 83 occasions in which it is required (cf. Figure 5 in appendix). This is true of anaphoric, but primarily of expletive use, as the latter occurs more frequently in the types of task given. As clitic forms generally are learned quite late (cf. Young-Scholten 1993), it is of note that numerous clitic forms can be found in Franca's utterances towards the end of the recording period. However, it seems that they primarily occur in formulaic expressions (in the formula gibt's from the 32 nd month on). 42 (25)NNS:

ja danach-eh (gib)/eh gib-s/eh *krach-e yes afterwards there's trouble zwischen + eh + poliziste between police und + eh arbeitlo/ci/arbeitlosen and unemployed 'Yes afterwards there's trouble between the police and the unemployed' (NNS Franca, 35 th month, Charlie Chaplin retelling)

Reference to the subject in German L2

39

5.2.4. der, die and das In the Franca data there is also anaphoric use of the pronouns der, die and das: there are 3 occurrences of the anaphoric der from the 26 th month, 43 19 of die from the 14th month, and 78 of das from the 11th month (cf. Table 2). It is evident that the pronoun das is used earlier and more frequently than the other two. (26)NNS:

er macht (auch) auf mein bild etwas he makes (also) on my picture something aber das gefällt mir nicht but that please me not 'He also does something to my picture, but I don't like that' (NNS Franca, 32 nd month, free conversation)

However, das does not seem to have the same significance for Franca as for the Heidelberg learners observed by Klein and Rieck (1982), for whom it was the first and most important device for anaphoric reference.

5.3. Subject

omissions

The analysis of the acquisition of explicit lexical forms of reference in Franca's variety has shown that: a) pronominal devices are acquired very early, b) anaphoric devices are acquired after deictic ones, and c) learner-specific subject omissions can be observed from the outset, and continue to occur for a long period of time. Some of these learner-specific subject omissions occur even when the corresponding pronouns are already used in different contexts. At the same time, an increasing number of subject omissions corresponding to target language usage occur (cf. Figure 6). When non-target-language-like subject omissions occur in the varieties of learners with a pro-drop source language, it would be reasonable to assume that this is caused by transfer. However, when Franca's recordings from the first cycle and her narratives from the 26 th and 35 th month are considered, it appears that the learner-specific subject omissions cannot be attributed to one single reason (i.e., transfer). In the following, the contexts of omission will be examined in closer detail.

40

Bernt Ahrenholz

a) Omission of es The pronoun es is omitted particularly often. However, it is important to differentiate between the following types of omission: -

Subject omission of the expletive es. The use of es is obligatory here; its function is purely morphological and syntactic (cf. Eisenberg 1999: 174). It occurs, for example, with verbs which state the existence of something. In the present corpus, the verb sein is affected particularly often (about 70 omissions of the expletive es occur in utterances with copula):

(27)NNS:

und so ist 0 [es] warm and so is 0 [it] warm ' A n d that is why it is w a r m ' ( N N S Franca, 8 th month, narrative)

Subject omissions of the (syntactical) es in the function of a formal placeholder. In these cases, es takes the place of the subject in the prefield, but is dropped when the prefield is occupied by an other element. Franca omits es irrespective of this condition: (28) N N S :

0 [es] sind viel leute eh wie/ wie 0 are many people like 'There are many people like m e ' ( N N S Franca, 14th month, conversation)

ichΛ I

During the first cycle, about 30% of all subject omissions involve the expletive and syntactical es. It would thus seem reasonable to treat these uses of es as a distinct learning problem, as the expletive or syntactical es is a semantical ly empty, grammatical subject. 44 -

Furthermore, in some cases an anaphoric reference to es (or das) is omitted.

(29) EXP:

NNS:

Imeinst du es war think you it was + deiner meinung in your opinion 0 [es/das] kann/ kann 0 [it/that] can

richtig right

sein be

was sie what she

richtig right

getan done

hat? has?

Reference to the subject in German L2

41

'Do you think it was right, what she did? in your opinion? May be right' (NNS Franca, 20 th month, expression of opinion) b) Subject Omissions of Deictic Pronouns Ich and du are used in most utterances with a deictic subject (see above). Nevertheless, learner-specific subject omissions do occur in some of the contexts in question. This is especially the case for deictic references with the article das, but also for the pronoun ich. (30) NNS:

0 [das] ist eine nee kiiche *soggiorno* 0 [this] is a no kitchen *living-room* 'That is not kitchen, it's a living room' (NNS Franca, 5th month, description of picture)

(31) NNS:

halp-e oktober eh komm 0 [ich] zurück half October come 0 [I] back Ί come back in the middle of October' (NNS Franca, 4th month, conversation)

Such omissions of ich can be attributed to the fact that the production of the pronoun is not yet fully automatic. With respect to the referential movement, omissions occur in the context of both "maintained" and "reintroduced" information. In a later stage of acquisition, however, subject omissions of ich can be interpreted as formulaic expressions also found in spoken German (weiss nich). Note, however, that as shown in example (32), the syntactic constraints of these forms are not necessarily taken into account: the use of aber 'but' does not allow the formulaic expression featuring the omission of ich in this case. (32) NNS:

aber 0 [ich] weiss nich but 0 [I] know not 'But I don't know' (NNS Franca, 14th month, problem solving)

c) Subject Omissions with Anaphoric Reference Many subject omissions occur with third person pronouns used in anaphoric contexts. This correlates with the later acquisition of anaphoric personal pro-

42

Bernt Ahrenholz

nouns. Correspondingly, the number of learner-specific subject omissions decreases over the course of acquisition (cf. Figure 6). (33)NNS:

aber hinter/ hinter der tram es gibt but behind the tram there is ein ande@re mann@ der hatte verschlaft another man he had overslep/ ehm_ + ver/ äh (schm/) geschlafenA slept und da hat-e 0 nicht bemerkt was passiert ist and there had 0 not noticed what happened is 'But there's another man who had been sleeping behind the tram, and had not noticed what happened' (NNS Franca, 26 th month, film retelling Harald Lloyd)

Subject omissions with anaphorical reference occur in the following contexts: - Subject omissions when the prefield is occupied by adverbs In approximately 50 of the 283 cases of learner-specific subject omissions analysed, the prefield is occupied by adverbs or complex adverbial expressions. Many of the cases in question involve constructions with und dann 'and then', und jetzt 'and now' and und so 'and therefore'. It might be assumed that Franca is experiencing difficulties in understanding the difference between this kind of "und + adverbial" utterance and co-ordinated utterances with und which do allow subject ellipsis in the case of referential identity. 43 However, no clear connection between the development of the syntax and subject omissions can be determined, contrary to the assumptions of Pienemann (1998) and Clahsen, Meisel and Pienemann (1983). (34) NNS:

*dunque* ein mannA ein morder ist-e well a man a murderer is ist-e + (h) eh gekommen eh in erste raumAeh a/ ehm is come in first room (h)at-e + kaputtΛ ehm (h)ate *va beh* kaputt has broke has okay broken fensterΛ un/ unt-e + at gefunden/ gefang/ gefunden window downstairs has found eh das mannA in erste raum_ unt-e\ the man in first room downstairs ehm

Reference to the subject in German L2

43

und dann-eA (h)at-e 0 [er] gemo/ gemort/ gemort'" and then has murdered eh mann-e mit-e eine stuhlΛ man with a chair 'Well a man, a murderer, came into the first room / broke the window downstairs / and then [he] murdered a man with a chair' (NNS Franca, 6th month, narrative dolls house) - Subject omissions with subordinate clauses Learner-specific subject omissions sometimes occur in conjunction with subordinate clauses. This applies to subordinate clauses placed after the main clause, as in example (35): (35) NNS:

ich glaub-ehm mein vater i/ is nich + eh zufrieden I believe my father is not satisfied EXP: m h m h NNS: weil 0 [er] immer + in-eh in alle leben hat immer because [he] always in in all life has always ehm + ehm ge/ e/ gearbeitet viel ier ier worked much here here 'Don't think that my father is very satisfied because he spent all his life here working really hard' (NNS Franca, 11th month, free conversation)

This also occurs in cases where subordinate clauses or infinitive constructions precede the main clause, as shown in the following example: (36) NNS:

aber eh fur gehen in eh/ehm neue stadt-e but for go in new town muss-e 0 [er] geh +2+nah eine/ein eh lange flussA must 0 [he] go near a long river 'But to go to the new town he has to go near to a long river (walk along a river)' (NNS Franca, 8th month, narrative) Here again, no regularity in the learner variety is clearly discernible: on the one hand, there are subject omissions in the context of subordinate clauses (16%, n=44 of all subject omissions). On the other hand, there are many constructions with subordinate clauses in which the subject is referred to explicitly, even in the case of maintained information.

44

Bernt Ahrenholz

d) Other Subject Omissions Although there seem to be particular contexts in which subjects are more likely to be omitted (about one-third of all cases of non-target-language-like subject omissions involve the expletive or grammatical es, about half occur in the context of the copula, some in the context of adverbs in the prefield, some in formulaic speech, and some in subordinate clauses), there are also occurrences of subject omissions without any facilitating surface conditions, as shown in the following example with maintained reference: (37) NNS:

mein vater ehm nich arbeitet mehr + my father not works more A e/ 0 [er] ist in *pensione * 0 is in Retirement* 'My father doesn't work any more / (he) is retired' (NNS Franca, 11th month, conversation)

Here, as in other examples shown above, possible transfer is facilitated by the conditions of referential movement. e) Target-Language-Like Subject Omission In addition to learner-specific subject omission, target-language-like subject omission occurs at a very early stage.46 The number of these subject omissions increases from the 8th month onwards, and they feature particularly often in narratives in the 14th and 26th month (cf. Figure 6). They occur parallel to the continuing learner-specific subject omissions and the use of pronouns. aber eine Jrau + %sieht-e + ehm die mädchenA-eh but a woman sees the girl und-eh eh 0 [sie] sagt-eeine: polizist and [she] tells a police officer 'But a woman sees the girl and tells a police officer' (NNS Franca, 14th month, Charlie Chaplin retelling) The increase in subject omissions which correspond to target language use (cf. Figure 6) first occurs in very short utterances (und sagt, und fährt weg, und geht da). With time the utterances in question become more complex. They occur mainly in co-ordinated clauses and in conjunction with the use of modal verbs (cf. Ahrenholz 1997,2000). But even if one can observe a development in the use of target-language-like subject omissions, in the Franca data there is (38) NNS:

Reference to the subject in German L2

45

a very frequent use of explicit reference to the subject for a long time, while in the productions of the German native speakers there is a significantly greater use of subject omissions (cf., e.g., example 8). 47

6. S u m m a r y and conclusion With respect to the acquisition of linguistic devices for reference to persons and objects in the function of subject, it has been shown that the learner acquires a comprehensive and differentiated repertory of lexical referential devices. In the learner variety analysed, NPs, personal pronouns, learner-specific and target-language-like subject omissions all occur, but personal pronouns emerge as being of primary importance for this domain of reference. In Franca's variety there is a clear difference with respect to the acquisition of deictic and anaphoric pronouns. Deictic pronouns, especially ich, but also du, are acquired very early, whereas anaphoric pronouns develop much later. This confirms what has already been demonstrated for other learner varieties (Klein and Rieck 1982). The pronouns ich and du are used from the beginning of the period of observation. Ich in particular shows a high number of occurrences, and is only rarely omitted. This, too, is in line with other research. The late acquisition of the anaphoric subject pronouns seems to be substituted first by subject omission and — to a lesser degree — by an overextended use of NPs. If we were to assume anaphoric pronouns to be more marked than NPs, as suggested by Lallemann (1993), a higher frequency of N P usage would be expected. Apart from personal pronouns, the articles das, die, and der are also used pronominally. This confirms the H P D (HPD 1977; Klein and Rieck 1982) findings that das is an important anaphorical device, although the pronominal use of these articles is not as widespread as in the H P D data. The use of the pronoun es, and especially the expletive and syntactical use of es, appears to present a particular learning problem. This is probably due to the fact that it often has only a syntactic function and is semantical ly empty. From this point of view, the inclusion of the expletive use of es in experiments on zero anaphora (as practised by some researchers) would seem rather problematic. All pronouns are used in correspondence with their meaning in the target language, with the exception of sie, which is occasionally used to refer generally to imaginary holders of office, a context in which native speakers would prefer the pronoun die or a passive construction.

46

Bernt Ahrenholz

Learner-specific subject omissions are characteristic of the early period of acquisition. A decline in their occurrence is particularly apparent when anaphoric pronouns have been acquired. However, it is characteristic of the acquisition process that even after the corresponding pronouns have been acquired, subject omissions continue to occur for a long time. This does not appear to represent temporary backsliding as much as a fundamental instability in the domain of subject reference. Indeed, the results of other studies confirm this assumption. 4 8 Certain surface structures seem to favour subject omission; this applies especially to cases in which the copula is used, or when the prefield is occupied by adverbial expressions. Even here, however, the learner variety is not consistent, as subject omission and explicit reference exist side by side. In sum, a three-phase structure emerges in the development of the domain of personal reference, as has also been shown for the acquisition of modal constructions (Ahrenholz 1998, 2000) and personal reference in other target languages (Chini 1998): After an early period of acquisition characterised by the lack of certain linguistic devices and the application of learner-specific solutions, there is a period of explicit reference; finally, in a third phase, the learner further acquires the possibilities provided by the principle of referential movement, and the power of implicitly maintained reference in the target language. This acquisition process seems to be steered by a combination of different factors: the function of the pronoun, transfer, and some surface conditions. As shown by Klein and Rieck (1982) and discussed by Klein (1990), and as found in the current study, the pronoun's functional complexity in that it serves as a deictic, anaphoric, or grammatical device seems to play the major role in the process of acquisition in the sense that it will be realised earlier and more consequently in some functions than in others. Studies that try to explain the acquisition of pronouns merely with reference to typological differences between source and target language such as the pro-drop parameter, or by looking at grammatical function of the pronouns only are therefore bound to overlook some of the more critical factors guiding the process. Transfer does seem to be the main reason for instability in the pronoun system of the learner variety, however, and the surface conditions outlined above seem to facilitate the influence of L I . Contrary to what might be expected on the basis of the ZISA results, however, there is no indication that syntactical development and the occupation of the prefield by adverbs are the main triggers for the omission or realisation of an overt subject.

Reference to the subject in German L2

47

Notes 1. The others are predicate, including objects and persons affected by the actions, time, space and modality; cf. Klein and von Stutterheim (1987) and von Stutterheim and Klein (1989); for a discussion of the term "reference" see von Stutterheim (1997) in particular. 2. According to Eisenberg (1999: 44), central categories for the subject function are nouns, nominal groups in the nominative, infinitives with zu, indirect interrogatives and conjunctional subordinate clauses. Hentschel and Weydt (1990: 318) mention nouns, pronouns, numerals, adjectives, infinitives used as nouns, infinitives with zu and subject clauses. 3. Cf. Klein and Rieck (1982). 4. This is not the case for children, who also refer to themselves and to others by name. Furthermore, as their deictic system still has to develop, children sometimes confuse du and ich (cf. Klein 1990; Slobin 1985). Corresponding phenomena have not been observed in adult language acquisition, but could perhaps occur as slips of the tongue. 5. These exceptions refer only to the spoken language. In some special kinds of texts, i.e., telegrams or advertising copy, subject omissions occur also more frequently, cf. Zifonun et al. (1997) on so-called "structural ellipses". 6. In the German terminology, prefield indicates the position in front of the finite verb in assertions. It can be occupied by an element such as the subject or an adverbial or subordinate clause; cf. Zifonun et al. (1997: 414ff.) and Auer (1993). In his study of verbs at the beginning of sentences in spoken German, Auer indicates that corresponding subject ellipsis in the plural is generally classified as unacceptable. Cf. also the detailed presentation in Fries (1988). 7. ' 0 ' indicates the omission of the subject (whether this is target-language-like or not). If necessary, an explanation is given in brackets (e.g. [ich]). 8. For constructions with subject omissions of the second person, cf. also Zifonun et al. (1997) and Auer (1993). 9. Another important difference is the existence of clitic pronouns. However, whether and to what extent this influences the acquisition of non-pro-drop languages is beyond the scope of this study. 10. A general overview of subject references in Italian can be found in Schwarze (1995) and Cordin and Calabrese (1988); a contrastive presentation in Koller (1983). 11. Although Italian is a pro-drop language, personal pronouns are used relatively frequently. According to the LIP (Lessico di frequenza dell 'italiano parlato, De Mauro et al. 1993) the forms io and tu are ranked 34th and 78th respectively in the list of most frequently occurring lexemes (noi 56th, voi 101st). In comparison, ich and du rank 6th and 21st respectively for the German native speakers in the distinctly smaller P-MoLL corpus. 12. NS = native speaker, NNS = non-native speaker, EXP = experimenter; cf. the list of transcription conventions in the appendix. Examples with no named speaker have been invented to illustrate a phenomenon.

48

Bernt Ahrenholz

13, Only options which are relevant to the varieties under investigation are mentioned here. Zero anaphora with infinitive constructions and i/ass-phrases are, for example, not discussed. 14. Similarly, the pronominal use of the negator kein (keiner, keine, keins). 15, "Le forme soggettive vengono sottintese quando e evidente di chi si parli, p.es.:... Ho invitato anche tua madre, ma non ha voluto accettare (sott, essa)"; (The subject remains implicit if it is clear who one is talking about, e.g., (I) did invite your mother too, but (she) didn't accept.); Regula and Jernej (1975: 141). 16 Cf. Cordin and Calabrese (1998) and Chini (1999). In the film retellings investigated by Chini, personal pronouns were rarely used with maintained references; the most frequent device was use of zero anaphora. The LIP lists a lower frequency of personal pronoun use for the third person than for the first or second person; nevertheless, lui {he) and lei {she) are among the 100 most frequently used lexemes {lui ranks 81st, lei 100th, loro 212 th ). The frequency of the demonstrative pronouns questo (this) (30 th ) and quello {that) (33rd) in the LIP corpus is striking. 17 Cf. von Stutterheim 1997; Ahrenholz 1998. 18 In the sense of Klein & von Stutterheim, cf. Footnote 1. 19 Cf. also the description in Ulbricht (1987) and the very detailed description in Fries (1988), who points out that no subject ellipsis is possible with the expletive For a detailed description of the film, see Klein & Perdue (1992: 8). 21 It is not necessary to chain every element with und. There are also connected utterances without an explicit connector (example 8). However, the last utterance does have to be joined by a connector, in general by und, the sequence of events would otherwise seem open, and the hearer would expect the speaker to continue. After (7") jetzt nimmst du die aktentasche / 0 sehst zu dem grünschwarzen stuhl, an explicit connector is expected: und 0 stellst die aktentasche links neben dich. 2 2 , There are a number of studies with other target languages, mostly English (Davies 1996; Flynn 1991; Fuller and Gundel 1987; Gundel and Tarone 1992; Hilles 1986; Munoz 1995; White 1985; Williams 1989) or Spanish (Liceras and Diaz 1999); many of them in a UG framework. 23, Diehl et al. (2000) conducted an extensive study of French-speaking pupils in Switzerland, who learned German at school (DiGS-corpus). Their data, which was based on written productions, showed that the informants did not have problems with subject pronouns, and that non-target-like subject omissions were almost nonexistent (237 ff.). 24, The informants Janka and Urzula did not use third person pronouns until the 23rd and 31st month respectively; only the informant Sascha used these pronouns at a very early stage (Dittmar and Skiba 1992). 25, In the data gathered from the Australian learner, there were also few omissions of pronouns (and counted together with the omitted copula); cf Pienemann (1998: 125). Because the learner acquired German exclusively in a tutored way, it is difficult to decide whether the acquisition of these pronouns is due to the fact that the 20,

Reference to the subject in German L2

49

source language is non-pro-drop or to the teaching environment. The same holds for the French-speaking pupils in the DiGS study (Diehl et al. 2000). 26. The learners in the DiGS project (Diehl et al. 2000) also restricted their use to subject pronouns for a long time; object forms were rarely used. 27. Furthermore, reference is made to examples illustrating the use of pronouns in German and Italian native speech. The examples of the German native speakers stem from the P-MoLL corpus, those of the Italian speakers from the P-MoLL corpus and the Ahrenholz IDI corpus (instructions in German and Italian), cf. Ahrenholz (1998a). 28. Occurrences not included in the tables and figures are repetitions, echoes and broken off utterances. 29. The transcriptions of the learners' utterances follow the conventions used in the PMoLL project (Dittmar et al. 1990); these are listed in the appendix. 30. Similarly, the learners studied by Klein and Rieck (1982) and Clahsen, Meisel and Pienemann (1983) also acquired ich relatively early, again not realizing it in all obligatory contexts. 31. The differing phases of acquisition for deictic and anaphoric pronouns also seem to undermine the idea that word order phenomena are decisive for the process of acquisition. 32. Where weiss nich is concerned, the Italian non so is first dominant in Franca's variety. In early recordings, it is also occasionally expressed in German as ich nich weiss-, later on (from the 14th month) it occurs very frequently as weiss nich. From this perspective, the question arises of how far early occurrences of nicht weiss and M'ohin nicht weiss are to be interpreted as transfer phenomena or as copies of corresponding formulaic expressions from the spoken language, or whether transfer and input act together here. 33. There are also some very rare cases (not considered here) in which du is used as an object pronoun: nah du (=neben dir 'near to you'); Franca, 6th month, 'ashtray' instruction). 34. Here, as defined by von Stutterheim and Klein (1989), side structures are all units of utterance which do not respond to the quaestio (cf. von Stutterheim 1997). 35. In the instructional discourses analysed, the domain of reference "persons and objects in the lunction of subject" is part of the topic in main structure utterances, and generally has the status of "maintained" information in the referential movement. On the one hand, this condition leads to a frequent use of the second person singular, on the other hand, the subject reference can be dropped if structural conditions allow, as in example (7). The difference between the instructions realised by non-native speakers and native speakers is that German native speakers make more use of the power of maintained information, and drop explicit references in co-ordinated structures more often than non-natives do (cf. Ahrenholz 1998,2000). 36. However, there are hardly any speech occasions in which the pronoun could have been used. The comparable corpus of native speakers also includes only one occurrence of ihr as a pronoun in the nominative (compared to 14 occurrences as possessive and 15 occurrences as a dative pronoun).

50

Bernt Ahrenholz

37. In addition, das (and die) are used as a pronoun with respect to object references. 38. This pattern may be due to the setting of the interviews and experiments, in which there was very little opportunity to use er (he) or sie (she, they) deictically. 39. It is striking that there are a number of self-corrections with respect to the choice of pronouns. On the basis of the present data, however, it remains open whether these are slips of the tongue or acquisition phenomena. 40. Difficulties with grammatical gender, which are of course present in the data, are of little importance in the context of pronouns, because nearly all of the pronouns used here refer to persons (biological gender). 41. Cf. Eisenberg 1994: 194f. Such formulaic expressions have also been observed by Young-Scholten in the acquisition of German as a second language, and in the acquisition of other lan42. guages by, e.g., Bartning (1997), who studied the acquisition of c'est. Cf. footnotes to the Tables. The articles are also used as relative pronouns (59 occurrences of die, 15 of das). It is not possible to judge whether the function is 43. relative or anaphoric in any particular case. Cf. Eisenberg (1999: 174). For this grammatical phenomenon and for a description of this learning problem in the learner language of Mexican students cf. Rail and 44. Rail (1988). To verity this interpretation, which was based mainly on the data from the first 20 months of recordings, all 175 occurrences o f j e t z t or dann in the prefield were ana45. lysed. The results show that the effect of these constructions is limited. The subject is omitted in only 25 of the 175 utterances; 87 utterances have a subject but no inversion; and Franca uses inversion in 63 cases after the 14th month. Interestingly, no connection is discernible between the development of syntax, inversion, and subject omission. 46, Whether or not the distinction between target-language-like and non-targetlanguage-like is valid from the learner's point of view is open to question. For the purposes of the present analysis, however, it seems appropriate to identify the utterances in which the learner variety syntax corresponds to the target language. 47, This tendency towards greater explicitness has also been shown in other contexts; cf. Ahrenholz (1998a, 2000) for modal devices, Chini (1998) for personal reference, and Munoz (1995) for Spanish learners of English. 48, Cf. details in the HPD (1977), ZISA (Clahsen, Meisel and Pienemann 1983) and Pienemann (1998). 49, In the entire corpus, there are 16 occurrences of ihr, 15 of these are as possessive pronouns. 50, The lexeme der occurs 334 times in the entire corpus. It is mainly used as an article from the 5th month, but there are 19 occurrences as a relative pronoun from the 21st month and only 3 occurrences as an anaphoric pronoun (cf. Table 2). 51 The lexeme die occurs a total of 966 times in the corpus. It is used as an definite article from the very first recording (885 occurrences), and as a relative pronoun (48 occurrences) and anaphoric pronoun (19 occurrences) from the 14th month. It only appears to be used deictically on 10 occasions, but it is not always entirely

Reference to the subject in German L2

51

clear whether anaphoric or deictic use is intended. Finally we have 4 other usages of the pronoun die. 52. In total, there are approx. 464 occurrences of das. It is used as an article in 203 cases, as a relative pronoun in 13 cases, and as an anaphoric pronoun in 78 cases (plus 37 other usages). The column "das + pres(entational)." has been included to distinguish between deictic use in "presentationals" such as Das ist ein Stuhl, which are used to identify objects and persons, and other deictic usages. 53. Includes singular and plural usages of sie (see also Fig. 4). In addition to occurrences which refer to the subject, there are also a few which refer to the object. These are not included here, however. 54. In total, there are 147 occurrences of es. Most cases involve the expletive or grammatical es (see also Fig. 5).

References Ahrenholz, Bernt 1997 Explicit and implicit reference in learner varieties. In: Lourdes Diaz and Carmen Perez (eds.), Views on the Acquisition and Use of a Second Language. Proceedings of the Eurosla-7 conference, 265-277. Barcelona: Universität Pompeu Fabra. Ahrenholz, Bernt 1998a Modalität und Diskurs — Instruktionen auf Deutsch und Italienisch. Eine Untersuchung zum Zweitspracherwerb und zur Textlinguistik. Tübingen: Stauffenburg. Ahrenholz, Bernt 1998b Rpo, zero anaphora, and clause combining. Paper presented at the ESF Conference 'The Structure of Learner Language,' Acquafredda di Maratea, Italy. Ahrenholz, Bernt 2000 Modality and referential movement in instructional discourse: Comparing the production of Italian learners of German with native German and native Italian production. Studies in Second Language Acquisition 22,3:337-368. Ahrenholz, Bernt 2004 Verweise mit Demonstrativa. Grammatische Beschreibungen, gesprochene Sprache, Zweitspracherwerb und Vermittlung im Deutschals-Fremdsprache-Unterricht (Habilitationsschrift, Freie Universität Berlin). Auer, Peter 1993 Zur Verbspitzenstellung im Gesprochenen Deutsch. Deutsche Sprache 3: 193-222.

52

Bernt A hrenholz

Bartning, Inge 1997 Portrait of the advanced learner and his learner variety, especially in French L2. In: Lourdes Diaz and Carmen Perez (eds.), Views on the Acquisition and Use of a Second Language. Proceedings of the Eurosla 7 conference, 15-25. Barcelona: Universität Pompeu Fabra. Broeder, Peter 1991 Talking about People. A Midtiple Case Study on Adult Language Acquisition. Amsterdam: Swets & Zeitlinger. Broeder, Peter 1995 Acquisition of pronominal reference: A longitudinal perspective. Second Language Research 11,2: 178- 191. Chini, Marina 1998 Testualitä e mezzi referenziali concernenti la persona in narrazioni di italofoni e di apprendenti di italiano L2. Studi Italiani di Linguistica Teorica e Applicata 27,1: 153-181. Chini, Marina 1999 Riferimento personale e strutturazione di testi narrativi in italofoni e in apprendenti tedescofoni di italiano. In: Norbert Dittmar and Anna Giacalone Ramat (eds.), Grammatik und Diskurs / Grammatica e Discorso. Studi sulVAcquisizione dell'Italiano e del Tedesco /Studien zum Erwerb des Deutschen und des Italienischen. Tübingen: Stauffenburg. Clahsen, Harald and Upyong Hong 1995 Agreement and null subjects in German L2 development: New evidence from reaction-time experiments. Second Language Research 11,1:5787. Clahsen, Harald, Jürgen Meisel, and Manfred Pienemann 1983 Deutsch als Zweitsprache. Der Spracherwerb ausländischer Arbeiter. Tübingen: Narr. Clyne, Michael 1968 Zum Pidgin-Deutsch der Gastarbeiter. Zeitschrift für Mundartenforschung 35: 130-139. Cordin, Patriza and Andrea Calabrese 1988) I pronomi personali. In: Lorenzo Renzi (ed.), Grande GrammaticaItaliana di Consultazione, 1, 535-592. Bologna: il Mulino. Davies, William 1996 Morphological uniformity and the Null Subject Parameter in adult SLA. Studies in Second Language Acquisition 18, 4: 475-493. De Mauro, Tullio, Federico Mancini, Massimo Vedovelli and Miriam Voghera 1993 Lessico di Frequenza dell'Italiano Parlato. Milano: Etaslibri. Diehl, Erika, Helen Christensen, Sandra Leuenberger, Isabelle Pelvat and Therese Studer 2000 Grammatikunterricht: Alles für der Katz? Untersuchungen zum Zweitspracherwerb Deutsch. Tübingen: Niemeyer.

Reference to the subject in German L2

53

Dittmar, Norbert, Astrid Reich, Magdalena Schumacher, Romuald Skiba, and Heiner Terborg. 1990 Die Erlernung modaler Konzepte des Deutschen durch erwachsene polnische Migranten. Info DaF 17: 125-172. Dittmar, Norbert and Bernt Ahrenholz 1995 The acquisition of modal expressions and related grammatical means by an Italian learner of German in the course of 3 years of longitudinal observation. In: Anna Giacalone Ramat and Grazia Crocco Galeas (eds.), From Pragmatics to Syntax: Modality in Second Language Acquisition, 197-232. Tübingen: Narr. Dittmar, Norbert 1993 Proto-semantics and emergent grammars. In: Norbert Dittmar and Astrid Reich (eds.), Modality in Language Acquisition/Modalite et Acquisition des Langues, 213-233. Berlin: de Gruyter. Eisenberg, Peter 1999 Grundriß der deutschen Grammatik. Der Satz. Stuttgart: Metzler. Flynn, Suzanne 1991 Goverment-Binding: Parameter-setting in second language acquisition. In: Thomas Huebner and Charles Ferguson, (eds.), Crosscurrents in Second Language Acquisition and Linguistic Theories, 143-167. Amsterdam, Philadelphia: John Benjamins. Fries, Norbert 1988 Über das Null-Topik im Deutschen. Sprache und Pragmatik. Arbeitsberichte·, 19-49. Lund: Universität Lund. Fuller, Judith and Jeannette Gundel 1987 Topic-prominence in interlanguage. Language Learning 37, 1: 1-18. Gass, Susan and Usha Lakshmanan 1991 Accounting for interlanguage subject pronouns. Second Language Research 7,3:181-203. Gundel, Jeanette and Elaine Tarone 1992 Language transfer and the acquisition of pronominal anaphora. In: Susan Gass and Larry Selinker (eds.), Language Transfer in Language Learning (revised edition), 87-100. Amsterdam/Philadelphia: John Benjamins. Heidelberger Forschungsprojekt "Pidgin-Deutsch" 1977 Heidelberger Forschungsprojekt "Pidgin-Deutsch spanischer und italienischer Arbeiter in der Bundesrepublik": Die ungesteuerte Erlernung des Deutschen durch spanische und italienische Arbeiter. Eine soziologische Untersuchung. Osnabrücker: Osnabrücker Beiträge zur Sprachtheorie, Beiheft 2.

54

Bernt Ahrenholz

Hendriks, Henriette 1998 Reference to person and space in narrative discourse: A comparison of adult second language and child first language acquisition. StudiItaliani di Lingiästica Teorica e Applicata 27,1: 67-86. Hentschel, Elke and Harald Weydt 1990 Handbuch der deutschen Grammatik. Berlin, New York: de Gruyter. Hilles, Sharon 1986 Interlanguage and the pro-drop parameter. Second Language Research 2: 33-52. Hong, Upyong 1995 Null-Subjekte im Erst- und Zweitspracherwerb. Eine vergleichende Untersuchung im Rahmen der Prinzipien und Parametertheorie. Tübingen: Narr. Keim, Inken 1984 Untersuchungen zum Deutsch türkischer Arbeiter. (Forschungsberichte des Instituts für deutsche Sprache 50). Tübingen: Narr. Klein, Wolfgang and Bert-Olaf Rieck 1982 Der Erwerb der Personalpronomina im ungesteuerten Spracherwerb. Zeitschrift für Literaturwissenschaft und Linguistik 12,45: 35-71. Klein, Wolfgang and Christiane von Stutterheim 1987 Quaestio und referentielle Bewegung in Erzählungen. Linguistische Berichte 109: 163-183. Klein, Wolfgang 1990 A theory of language acquisition is not so easy. Studies in Second Language Acquisition 12: 219-231. Klein, Wolfgang and Mary Carroll 1992) The acquisition of German. In: Wolfgang Klein and Clive Perdue (eds.), Utterance structure: Developing grammars again, 123-188. Amsterdam: Benjamins. Klein, Wolfgang and Clive Perdue (eds.) 1992 Utterance structure: Developing grammars again. Amsterdam: Benjamins. Kohlmann, Ute 1997 Objektreferenzen im Kontext. Eine Untersuchung zum Zusammenhang von kommunikativer Aufgabe, Textstruktur und Objektreferenzen. Frankfurt: Lang. Koller, Erwin 1983 Zum Subjektpronomen aus kontrastiver Sicht: Deutsch-Italienisch. In: Maurizio Dardano, W. Ude Dressler and Gudrun Held (eds.), Parallela. Akten des 2. österreichisch-italienischen Linguistentreffens / Atti del 2 convegno italo-austriaco SLI, 304-315. Tübingen: Narr.

Reference to the subject in German L2

55

Kuhberg, Heinz 1990 Zum L2-Erwerb zweier elfjähriger Kinder mit Türkisch und Polnisch als Ausgangssprachen: Eine Longitudinalstudie unter besonderer Berücksichtigung kontrastivlinguistischer Gesichtspunkte. Deutsch Lernen 15,1: 25-43. Kuhs, Katharina 1989 Sozialpsychologische Faktoren im Zweitspracherwerb. Eine Untersuchung bei griechischen Migrantenkindern in der Bundesrepublik Deutschland. (Language Development 10). Tübingen: Narr. Lalleman, Josine 1993 A functional approach to markedness: Possible uses in the description and explanation of features in the Dutch of advanced second language learners. In: Bernhard Kettemann and Wilfried Wieden (eds.), Current Issues in European Second Language Acquisition Research, 29-46. Tübingen: Narr. Liceras, Juana and Lourdes Diaz 1999 Topic-drop versus pro-drop: null subjects and pronominal subjects in the Spanish L2 of Chinese, English, French, German and Japanese speakers. Second Language Research 15,1: 1-40. Müller, Heinrich 1993 Stärken und Schwächen vier microcomputerfahige Programme zur Textanalyse. In: Winfried Lenders (ed.), Computereinsatz in der Angewandten Linguistik. (Forum Angewandte Linguistik, Bd.25), 77-84. Munoz, Carmen 1995 Markedness and the acquisition of referential forms. The case of zero anaphora. Studies in Second Language Acquisition 17,4:517-527. Perdue, Clive (ed.) 1993a Adult Language Acquisition, Cross-linguistic Perspectives: Volume 1. Field methods. New York: Cambridge University Press. Perdue, Clive (ed.) 1993b Adult Language Acquisition, Cross-linguistic Perspectives: Volume 2. The results. New York: Cambridge University Press. Pienemann, Manfred 1981 Der Zweitspracherwerb ausländischer Arbeiterkinder. Bonn: Bovier Verlag. Pienemann, Manfred 1998 Language Processing and Second Language Development. Processibility Theory. Amsterdam, Philadelphia: Benjamins. Rail, Dietrich and Marlene Rail 1988 Das kommt uns spanisch vor: rätselhafte Pronomen. In: Pavica Mrazovic and Wolfgang Teubert (eds.), Valenzen im Kontrast. U. Engel zum 60. Geburtstag, 328-346. Heidelberg: Groos.

56

Bernt A hrenholz

Regula, Moritz and Jospip Jernej 1975 Grammatica Italiana Descrittiva su Basi Storiche e Psicologiche. Bern and München: Francke. Rieck, Bert-Olaf 1989 Natürlicher Zweitspracherwerb bei Arbeitsimmigranten. Eine Langzeituntersuchung. Frankfurt am Main: Peter Lang. Schumacher, Magdalena and Romuald Skiba 1992 Prädikative Ausdrücke in den Lernervarietäten einer polnischen Migrantin. Eine Longitudinalstudie. In: Linguistische Berichte 141: 371-400. 142: 451-476. Schwarze, Christoph 1995 Grammatik der italienischen Sprache, 2. revised edition. Tübingen: Niemeyer. Skiba, Romuald and Norbert Dittmar 1992) Pragmatic, semantic, and syntactic constraints and grammaticalization. Studies in Second Language Acquisition 14: 323-349. Slobin, Dan 1985 Crosslinguistic evidence for the language-making capacity. In: Dan Slobin (ed.), The Crosslinguistic Study of Language Acquisition, Vol. 2, 1157-1265. Hillsdale, NJ: Erlbaum. Stutterheim, Christiane von and Wolfgang Klein 1989 Referential movement in descriptive and narrative discourse. In Rainer Dietrich and Carl Friedrich Graumann (eds.), Language Processing in Social Context, 38-76. Amsterdam: Benjamins. Stutterheim, Christiane von 1997 Einige Prinzipien des Textaufbaus. Empirische Untersuchungen zur Produktion mündlicher Texte. Tübingen: Niemeyer. Ulbricht, Adelheid 1987 Subjektreduzierungen — eine Fehlerquelle bei Deutsch lernenden Ungarn. Deutsch als Fremdsprache 24,3: 172-176. Williams, Jessica 1989 Pronoun copies, pronominal anaphora and zero anaphora in second language production. In: Susan Gass, Carolyn Madden, Dennis Preston and Larry Selinker (eds), Variation in Second Language Acquisition. Vol. I: Discourse and Pragmatics, 153-189. Clevedon: Multilingual Matters. Young-Scholten, Martha 1993 The L2 acquisition of informal speech in German. In: Bernhard Kettemann and Wilfried Wieden (eds.), Current issues in European Second Language Acquisition Research, 111-124. Tübingen: Narr. Zifonun, Gisela, Ludger Hoffmann, and Bruno Strecker 1997 Grammatik der deutschen Sprache. Berlin, New York: de Gruyter.

Reference to the subject in German L2

Appendix Transcription Conventions Native speaker NS Non-native speaker NNS experimenter EXP + silence silence + seconds +n+ simultaneous speaking &XX& foreign language *xx* (xx) assumed lexeme

?xx?

@ (h)

/ Λ

57

questions lengthening laughing breathing cut off rising intonation falling intonation

Table 1. Franca, deictically used pronouns (absolute numbers) Sie der30 die31 wir ihr49 ich du

das'2

das + pres 1

th

72

7

5

th

54

6

1

2

th

26

10

0

1

th

13

1

1

6

9

9 m.

77

6

4

2

1

th

47

25

13

4

2

th

77

67

14

1

3

th

111

34

12

1

1

th

75

24

2

8

4

145

ο J

5

5

4

1

4 m. 5 m. 6 m. 8 m. th

11 m. 14 m. 18 m. 20 m. st

21 m. rd

m.

65

34

6

th

24 m.

115

5

7

th

145

33

23

25 m. th

4

6

8

2

3

1

1

m.

38

49

0

th

28 m.

27

11

3

st

68

25

2

26

31 m. nd

m.

102

72

6

th

34 m.

93

1

5

th

64

17

2

lh

77

48

1

th

39

9

3

1530

487

100

32

35 m. 36 m. 38 m. Total

1

1

5

1

26

2

3

2

4

3

7

17

5

4

3

2

5

1

4

12

1 1

36

0

10

66

67

58

Bernt Ahrenholz

Table 2. Franca, anaphorically used pronouns (absolute numbers)

sie'3

er

th

es

4 month

2

5th month

1

54

der

das

die

th

6 month 8th month th

2 1

2

th

7

4

th

16

4

9 month 11 month 14 month lh

1 1

2

1

18 month

9

2

20th month

24

12

6

2

1

month

4

4

2

1

24th month

13

25

j

3

4

3

3

2

3

3

4 12

1

11

2

8

st

21 month 23

rd

4

th

3

3

th

18

44

4

lh

26

25 month 26 month 28 month s

31 ' month

15

6

1

32nd month

1

5

2

th

3

8

1

th

18

14

1

th

15 15

1

1

194

139

19

34 month 35 month 36 month th

38 month Total:

8

2

1

6

3

8

1 2

4

19

78

Reference to the subject in German L2

59

Franca. The pronoun "ich" 100% 90%

80% 70% 60% 50% 40% 30%

20% 10% 0%

9 non so'

4th month

5th month

6th month

8th month

9th month

11th month

14th month

18th month

24

22

23

9

3

6

3

9

3

6

2

9

4

1

1

1

0

2

19

• IL omission of 'ich' S s u b . om. formulaic

20th month

26th month

35th month

5

1

0

0

1

0

0

0

13

14

2

4

S T L omission

2

0

2

0

4

1

1

1

0

0

0

• ich'

72

54

26

13

77

47

77

111

75

38

64

Figure 1. Franca. The pronoun 'ich'. 'non so' IL omission of 'ich' sub. om. formulaic TL omission ich

= use of the Italian formulaic form 'non so' (don't know) = learner-specific subject omission for the first person singular = use of formulaic subject omissions, especially 'weiss nich' (don't know) = subject omission corresponding to target language use = use of the pronoun 'ich'

60

Bernt Ahrenholz

Franca. The pronoun 'du' and verb morphology

Figure 2. Franca. The pronoun 'du' and verb morphology Vit/Vmiss = MV#= V# = V-en = hast; bist = MV-st/-s = V-st =

Use of Use of Use of Use of Use of Use of Use of

'du' in conjunction with an Italian verb or verb missing 'du' in conjunction with modal verbs with deviant morphology 'du' in conjunction with verbs with deviant morphology 'du' in conjunction with verbs with the ending '-en' 'du' in conjunction with the verbs 'hast' (have) and 'bist' (are) 'du' in conjunction with modal verbs with the ending '-.ν'or '-st' 'du' in conjunction with verbs with the ending '-st'

Reference to the subject in German L2

61

Franca. The pronoun 'er1 100% I

ρ

1Π

1

I

ι

I

80% 1 I I

I

äE

s

1

ää

1

!

1

1I

1 60% n

"

40% -

20% -

I Γ1

4

8

• TL omission • er'+verb

•

2

1

!

*

1

L 0% - l ! 4th 5th 6th month j rnonth 1 rnontlτ

Ξ IL omission

m

I 1

I

8th month

9th ιTionth

11th month

4

12

7

10

1

7 2

1 1th m nth

I Π 18 th mo nth

20th ηlonth

i

5

8

3

4

I

24th nontt ι

26th 34th month ) month

3 1

5

10

24

'

35th \ month

9

:»

22

3

2

42

8

13

I

1

Figure 3. Franca. The pronoun 'er'. Occurrences of the pronoun 'er'in absolute numbers

IL omission = TL omission = 'er' + verb =

learner-specific subject omission subject omission cooresponding to target language use use of the pronoun 'er' with verb

62

Bernt Ahrenholz Franca, "sie". 3rd P e r s o n Pronoun

Figure 4. Franca. 'Sie' 3 ,d person pronoun. IL omission 'sie' pi = IL omission 'sie' sgl = TL omission 'sie' pi = TL omission 'sie' sgl = 'sie' pi + verb = 'sie' sgl + verb =

learner-specific subject omission of the plural 'sie' learner-specific subject omission of the singular 'sie' subject omission of the plural 'sie' corresponding to target language use subject omission of the singular 'sie' corresponding to target language use use of the pronoun plural 'sie' with verb use of the pronoun singular 'sie' with verb

Reference to the subject in German L2

Franca. The pronoun "es"

Figure 5. Franca. The pronoun 'es'.

IL omission = TL omission =

learner-specific omission of 'es' omission of 'es' corresponding to target language use

64

Bernt Ahrenholz

Franca. Maintained reference third person

E3NP E3 3rd person pronoun

r

4th month

5th month

6th month

8th month

9th month

11th month

e

β

0

5

2

1

0

0

3 7

0

15

• TL subject omission

0

1

0

• iL subject omission

β

19

6

ι

14th month

18th month

20th month

3

7

2

3

3

0

4

10

7

3

24

17

4

6

17

1

2

20

5

8

17

10

4

8

10

1

Figure &. Franca. Maintained reference third person.

26th month

35th month

Reference to person in learner discourse Marina Chini 1. Introduction In order to become a competent speaker of an L2 one has to learn to use various referential expressions that are appropriate to the discourse context, to mutual background knowledge and, obviously, to L2 grammar, supplying the interlocutor with a clear temporal, spatial and personal frame and signalling possible changes to this frame. In everyday conversation, and especially in narrative discourse, personal reference plays an important role both at the ideational as well as at the textual, and possibly at the interpersonal level: animate referents are often text topics and carry out, as such, an essential cohesive function, apart from being of special relevance from the informational point of view. Appropriate referential movement in the personal domain requires a certain degree of L2 command and knowledge of L2 grammar, forms and discourse patterns related to the distribution of given and new information in the text. In order to take into account this important relationship between referential movement and text structure, and in line with the general approach of this project, we will adopt the Quaestio model as a framework for our text analysis. This provides a useful and language-adaptable interface between the conceptualisation and the formulation of a text (Levelt 1989), accounting both for textual coherence and cohesion and contributing to a better understanding of the choice of referential devices in texts. According to this model, a text can be seen as an answer to a specific question or quaestio that imposes constraints on the macrostructure of the text, on topic/focus distribution and on the form of its single sentences, including its referential elements. Formal consequences of the quaestio can differ from one language to another. Such a general framework fits well not only for native speaker texts but also for learners'; it can be supposed that, if they dispose of sufficient linguistic and textual competence, the latter will obey these constraints in a more or less target-language-like way. Within the above framework the present contribution examines the use of referential forms in post-basic varieties of learners who are supposed to have overcome the pragmatically oriented stage with its "non-finite utterance organisation" and to have access to a more grammaticalised ("finite") utterance organisation (Klein and Perdue 1992, 1997; Perdue 1993). Such a level of

66

Marina Chini

competence should enable them to carry out their task relying not only (or mainly) on basic and general procedures for organising information, but also on the more specific structuring devices of the target language, and to produce a discourse with a certain degree of cohesion (inter alia, by means of anaphoric chains). Our aim is to discover to what extent post-basic learners' discourse patterns, and particularly their use of anaphoric devices, conform to or diverge from native speaker patterns, in order to better understand language development beyond the most basic patterns in this area (which is one of the main topics of the SLV Project, see the Introduction to this volume), and to uncover possible learning difficulties along with their causes. Our research questions are the following: -

-

How do learners express and integrate information from the semantic domain of person when producing a coherent (narrative) text at a given time? In other words: what are learners' choices in referential movement with regard to the personal domain? How do their procedures change at different levels of competence? Which factors affect referential choices in post-basic to advanced learner varieties?

2. Earlier studies Studies on reference have adopted several perspectives: theoretical and philosophical, descriptive, typological, and, of course, acquisitional. It is not possible here to extensively review all this literature (cf. Andorno 2003; Sterelny 1994; Garrod and Sanford 1994 for rapid reviews); before concentrating on some acquisitional studies in this field it would be useful to remember some basic concepts and works of more general interest.

2.1. Reference Reference (Halliday and Hasan 1976) is a concept belonging primarily to the philosophical (Frege, Russel, Wittgenstein), semiotic (Ogden and Richards) and semantic tradition (for a synthesis, cf. Lyons 1977; Sterelny 1994). Simply put, reference is "the relationship which holds between words and things" (Lyons 1968: 404), or, better, the speaker himself refers to various (real or fictitious) entities by using some appropriate referential expressions (Lyons 1977: 177). Referential expressions are not only linguistic, but also, ontogenetically earlier, non-linguistic (i.e. pointing), in L2 as in LI (Atkinson 1979).

Reference to person in learner discourse

67

We cannot treat the non-linguistic means here, but we will concentrate on linguistic reference, and in particular reference to extralinguistic entities (= referents), mainly persons; deictic and textual references are neglected in the present study. Referents belong to one of the following five domains: persons and objects (= entities), times, places, actions-events, modalities. According to the Quaestio Model any utterance selects referents from these domains and integrates them into the proposition; within a text, this information shifts from one utterance to the next, producing the so-called "referential movement", in the various domains (von Stutterheim and Klein 1989: 44-48). Referential movement contributes both to textual coherence (at the semantic-pragmatic level) and cohesion (at the formal level, through grammatical and lexical means, like anaphors, markers of temporal, modal and local continuity, etc.: von Stutterheim and Klein 1989 1 ), and can have different surface translations in different languages. As for the personal domain, in narratives we often have reference to one or more specific persons (the narrator or the protagonist(s)), maintained through the utterances of the main structure. This "topic continuity" is an important structural feature of narrative discourse (von Stutterheim and Klein 1989). Reference to various entities is often made by means of NPs of different forms: definite and indefinite NPs, full and reduced pronouns, zero pronouns, zero anaphora. Typological works have analyzed the occurrence of these elements and the relationship between reference to the topic/ topicality/ coreference and textuality in several types of languages 2 . In particular a famous cross-linguistic quantitative study by Givon (1983) and associates has proposed the following scale of referring expressions in relation to the accessibility of the topic: (1)

Most continuous/accessible topic zero anaphora unstressed/bound pronoun or grammatical agreement stressed/independent pronoun Right-dislocated DEF-NPs neutral ordered DEF-NPs Left-dislocated DEF-NPs Y-moved Nps contrastive topicalisation cleft-focus constructions referential indefinite NPs Most discontinuous/inaccessible topic

68

Marina Chini

The basic principle underlying Givon's scale is iconicity: the most salient referential forms (referential indefinite NPs, cleft-focus constructions) are employed for the most disruptive/new topics, the least salient and formally light forms (pronouns, zero anaphora, grammatical agreement) for the most continuous ones. We can also assert that the means which are favoured for poorly accessible referents are explicit, in the sense that they include the noun of the referent (in various types of NPs), while the means which are employed for accessible referents tend to be more implicit, and pronominal in nature (stressed and unstressed pronouns, inflectional agreement, zero anaphora). According to other researchers various (textual, cognitive) factors seem to have consequences for the use of referential means 3 . Referential devices are normally more explicit in written formal texts, at the beginning of text units and for inanimates than in other cases. Anaphoric references to people, being in general more salient and topical, should then be more implicit than references to inanimates. The grammatical properties of individual languages also severely constrain the range and frequency of referential means and their textual use. In pro-drop languages like Italian, zero pronouns/grammatical agreement for referential continuity in subject position are the rule (Renzi 1988; Conte 1988; Berretta 1990a), whereas in non pro-drop languages like English and German they are possible only in precise syntactic conditions (coreference with the subject of the preceding conjoined clause). In particular, Italian uses all the means listed by Givon, with some preferences: verbal agreement and clitics are very often used for continuous topics respectively in subject and non-subject (mostly direct or indirect object) position, whereas stressed pronouns are mainly employed for emphatic and contrasted anaphors, so they are more marked and less frequent than in English or German. As for zero anaphors (null subjects of non-finite subordinate clauses), they are not so frequent as in Chinese or Japanese; lexical anaphors (definite NPs) are normally used for less continuous referents (right dislocated definite NPs codify more accessible topics than leftdislocated NPs; clefted NPs indicate the most discontinuous ones); for newly mentioned referents Italian typically relies on postverbal indefinite NPs 4 .

2.2. The acquisition of language specific rules in LI How are these language-specific rules - and more generally how are the linguistic means for referential movement - acquired? Literature in this field is abundant with regard to LI acquisition (cf. reviews in Min 1994: 34-56 and Hickmann 1995: 204-208). Works in this field adopt one or more of the fol-

Reference to person in learner discourse

69

lowing perspectives: psychological, syntactic and textual-discursive. We cannot treat them here extensively, but only mention those analyses and results which are especially relevant to our purposes. Research carried out from the psychological perspective has reported that deixis, being related to the immediate context of speaking, turns out to be the first kind of reference in child language, being more basic than anaphora (also the deictic use of personal pronouns is more basic than anaphoric use; Karmiloff-Smith 1979, 1981; Hickmann 1991, 1995). Syntactic-oriented research focuses on the acquisition of N P types, especially of pronouns, determiners or other local (numerals, demonstratives, classifiers) and global (clause structure variations, like subject-verb inversion) referential means 5 . Researchers do not agree upon the time of the acquisition of N P types; some scholars claim that it is relatively early (Brown 1973; MacWhinney and Bates 1978; Maratsos 1979; Bamberg 1987), others relatively late (Hickmann 1991; Karmiloff-Smith 1981; Power and Dal Martello 1986; Wigglesworth 1990) 6 . In fact some devices appear early, but are not immediately used in the same way as by adults. Crosslinguistically, children often make errors in using definite and indefinite articles, because they find it difficult to take into account the addressee's point of view (Brown 1973; Hickmann 1987). As for Italian L I , definite articles appear earlier than indefinites (at around two years of age or before) and are more widely used, sometimes as a simple N P marker (Cipriani et al. 1993), sometimes as a specificity marker (not always appropriately), indicating the speaker's familiarity with the referent, more than the listener's knowledge (Power and Dal Martello 1986). Italian third person clitic pronouns appear almost simultaneously with definite articles, while third person stressed pronouns (lui, lei) are (amazingly) relatively late (2;6-3;6; Pizzuto and Caselli 1992; cf. Chini 1995: 129-146 for a review). Children's use of clitic and stressed pronouns is initially often exoforical; at 4 years they also begin to be used anaphorically, as a typical reference maintaining device (Orsolini and Di Giacinto 1996; Orsolini et al. 1996). Furthermore Italian children employ a light grammatical referential device early on: zero subject pronouns (pro-drop) and verb-subject agreement 7 ; they also resort to full NPs as subjects, but use very few full pronominal subjects (Hyams 1986; Pizzuto and Caselli 1992 8 ; Caselli et al. 1993). On the basis of Italian L1 data compared with English data, generative scholars have proposed that the pronoun-dropping (pro-drop) option constitutes the initial, unmarked setting of the so-called pro-drop parameter (Hyams 1986). This could also have consequences for L2 acquisition: learning a nonpro drop (marked) L2 while having a pro-drop (unmarked) LI should be more

70

Marina Chini

difficult than the reversed situation (LI non-pro-drop, L2 pro-drop, as in our case). In fact, available data shows that English children begin with subjectless sentences, shifting to the use of pronominal subjects just before the acquisition of verbal morphology, when they "set the parameter" in conformity with its position in the fully-fledged language (cf. also Phinney 1987). Analyses of the development of reference tracking devices in LI adopting a textual perspective are the most interesting for our purposes. They often treat reference in narrative data (Karmiloff-Smith 1981; Bamberg 1986; Hickmann 1991, 1998; Hickmann et al. 1996; Wigglesworth 1990; Berman and Slobin 1994), finding universal trends in textual development, but also some language-specific differences. It has been reported that, for various languages (including Italian and German, cf. Orsolini and Di Giacinto 1996; Bamberg 1987), 4-year-old children already tend to introduce or re-introduce referents by means of full NPs and to maintain reference by weak forms (pronouns, person/number inflection on the verb). Some other studies report that children fully learn to mark newness only from 7 years on (Hickmann 1995; Hickmann et al. 1996). In general it can be said that the use of referring expressions in absence of mutual knowledge, and particularly means for referent introductions, are acquired late (Kail and Hickmann 1992), and that reference maintenance is less explicit with animate referents (especially with the main characters) than with inanimates (Karmiloff-Smith 1981; Bamberg 1987; Kail and Hickmann 1992). Furthermore it must be said that use of referential forms is both intertwined with children's development of narrative discourse competence (cf. Berman and Slobin 1994) 9 , which in adult learners should be already there (at least for LI), and affected by the system of devices that is available/preferred in the language they are acquiring 10 . In conclusion, LI data show a complex (and not always coherent) picture, with a clear developmental progression from deictical referential devices to anaphoric ones; from relying on the nonlinguistic context to resorting to discourse· internal referring expressions; from a general tendency to distinguish given from new referents (often on the speaker's knowledge basis) to the acquisition of language-specific local and (then) global devices for marking such a distinction; contributing more and more to textual cohesion in a language specific way. Various semantic, pragmatic and formal factors seem to affect the acquisition of referential means in the first language.

Reference to person in learner discourse

71

2.3. The acquisition of language specific rules in L2 In L2 acquisition (where cognitive difficulties should be lower) the tendency to distinguish new referents from old ones comes earlier than in L I , but the means employed by learners are not always (morphologically) correct, although often pragmatically appropriate (Hendriks 1998). Apart from deictic means and gestures, in early varieties reference is often made with bare nouns and names, or is left implicit, to be inferred from the context. Early varieties typically lack articles, an important means to distinguish definite NPs from indefinite ones; some anaphoric (and, of course, deictic) pronouns are present, but overextended (i.e. he in English L2, sie in German L2). In a successive stage, in order to better distinguish new topics from old ones, learners begin to use some articles (and/or demonstratives) to form definite and indefinite NPs, without initially inflecting them correctly (for gender, number and/or case, if necessary; Berretta 1990a; Chini 1995, for Italian; Klein and Perdue 1989, 1992; Dietrich 1982; Hendriks 1998 and Wegener 1995 for German L2 etc.). Some research reports that definite referential forms are acquired earlier than indefinite ones, in L2 as in LI (Dietrich 1982; Chaudron and Parker 1990; Broeder 1991: 127-150; Hendriks 1998), but there are also counterexamples (Pfaff 1987; Berruto et al. 1990)11. Some specific studies dealing also with personal reference are worth mentioning here. Referential devices in basic varieties (L2 German, French and English) are treated from the perspective of "learner's problem of arranging words" by Klein and Perdue (1989). They find three pragmatic principles playing a role in this regard: familiarity with the referent; introduction or maintenance of the referent; topic/focus structure and topic-focus transitions' 2 . In general in basic varieties new referents are in focus (end) position and are expressed by lexical means (indefinite NPs or bare nouns), while pronouns (also zero pronoun) refer to topical referents; demonstratives are used for transitions from focus to topic. Broeder's (1991) longitudinal study of two Turkish and two Moroccan adults learning Dutch gives a more detailed picture on the use of referential means in L2. His findings suggest that, unlike in LI acquisition, L2 learners are aware of the global structuring of information already in early stages, and that there are no strong developmental changes over time. Usually full NPs are used for the establishment of reference (but they are mostly definite, rather than indefinite). The effect of the thematic status of characters can be observed for maintenance and shift of reference: in both cases full NPs (bare nouns and definite full NPs) dominate for minor characters, while pronouns are frequent for the main character (for reference maintaining, zeros are also found; Broe-

72

Marina Chini

der 1991: 127-150)13. Only some constructions (NP dislocations and constituent fronting) point to a transfer of LI information structure in these Dutch L2 data. Other research work with an ESF background evidentiate problems (i.e. omissions) with local marking like articles (i.e. in French; Veronique 1989, 1990), or with pronominal reference (Breeder et al. 1988: 86-113; Giacobbe 1992) and a tendency to rely mainly on word order at first (given information preverbal; new information postverbal; cf. Dietrich 1982 for early German L2). The phenomenon is also analysed in a more syntactically oriented approach either from a generative or a functional perspective, in several target languages (English L2: i.e. White 1985; Chaudron and Parker 1990; Munoz 1995; Italian L2: Valentini 1992; Spanish L2: Liceras 1989). In the Government and Binding framework, non target-like omission of subject pronouns in English L2 has been underlined and interpreted in different ways: as evidence for the unmarked status of the pro-drop setting of the parameter (Phinney 1987) or as evidence for the influence of L i s like Italian, Spanish, Chinese (White 198514). The two hypotheses (UG and transfer) are not incompatible, according to Liceras (1989). Her study of Spanish L2 by French and English learners seems to provide evidence for the unmarked status of the pro-drop option: learners with different proficiency levels and with a non-pro-drop LI accept Spanish null subjects without difficulty and seldom transfer the non-pro-drop option of their LI to L2. Recent studies within the generative model propose other kinds of explanations, which also offer alternatives to parameter (re)setting' 5 . Whatever principled explanation ofthe data is given, this kind of framework does not seem suitable for explaining the whole pattern of omissions and uses of subject pronouns in learners' texts, which is very much discourse-related (and not only parameter- or grammar-related), as we will see. From a different, more functional perspective, Chaudron and Parker (1990) offer a markedness 16 interpretation of NP acquisition by Japanese learners of English L2: with increasing proficiency, forms which are considered more marked (definite, also left-dislocated, NPs and existential indefinite NPs) should be more widely used, and the least marked forms, such as pronouns and bare nouns, less used. The data shows that pronouns are more frequent in English L2 than LI, in all contexts and level groups, while English natives use more full (definite and indefinite) NPs than learners; but no explanation is given for the lack of zero anaphora in English L2, a form the authors classify as structurally unmarked and that, strangely enough, these learners never use in L2, although they often use it in their LI. Chaudron and Parker's markedness interpretation is in our view debatable: are structurally light referential means, like pronouns and zero anaphora, necessarily less marked than full

Reference to person in learner discourse

73

NPs, at least from the acquisitional and cognitive point of view? We will come back to this issue in the conclusion. Replicating Chaudron and Parker's study with Spanish learners of English, Carmen Munoz (1995) also brings some of their claims into question. She too finds an overuse of pronouns in contexts of topic continuity and few instances of zero anaphora in her subjects: only intermediate and high level learners use zero anaphora, and they always do it correctly, for current topics and in conjoined clauses. The rare occurrence of zero anaphora contradicts Chaudron and Parker's assumption about its unmarked status. Munoz explains her findings by appealing to the clarity of the L2 rule (obligatoriness of the subject in English), to the role of its frequency in the input, to formal instruction insisting on grammar (learners followed an intensive summer program) and to the controlled (written) task. Polio (1995) as well, looking at English and Japanese learners of Chinese, finds an underuse of zero pronouns and an overuse of lexical pronouns in L2 data. LI transfer cannot explain these findings (zero anaphora is widely used in Japanese), but the search for clarity, the salience of these forms and their frequency in the input (teacher talk) partly can. On the other hand, a recent study on Chinese varieties of English learners (Jin 1994) underlines the role of LI typology, outlining an initial systematic process of transferring English subject-prominent features to Chinese L2, which gives rise to few null subjects 17 . Hence, omission of the subject pronoun does not seem to be as general a phenomenon of learner varieties as was claimed when studies on English L2 were prevalent. So Jin brings into question the topic-prominent character of the interlanguage maintained by Fuller and Gundel (1987). However, there are also cases where zero anaphora is used more frequently by non-native speakers than by native speakers, i.e. in the case of Singaporean speakers of English (LI Malay, Tamil and several dialects of Chinese) and of English learners with different L i s (Mandarin, Cantonese, Russian, Kikongo etc., mostly pro-drop Lis; Williams 1988). The general discourse function for zero anaphora is similar in learners and native speakers (marking continuous, thematic referents), but these non-native speakers use it in contexts where the referential distance from the antecedent is greater than in contexts where native speakers use it, thus giving rise to ambiguities. According to Williams, two opposing forces explain deviations from target language norms: economy (minimalisation of redundancy) and the search for hyperclarity. Finally it is worth mentioning studies which underline the discourse and textual function of referential forms, without neglecting the correlation between their use and morphological development. A compensatory use of subject pronouns has been attested in Moroccon Arabic L2 narrations of an Eng-

74

Marina Chini

lish learner with deficient verbal morphology (Fakhri 1984), and also in learners of Italian (Berretta 1990b; Valentini 1992) and German (Meisel 1991) with different Lis. In the Italian L2 of two young (12 and 17 year old) Chinese learners, the use of subject pronouns is more frequent in early phases when verbal morphology is not clearly present (lui detto lit. 'he said PAST PART' instead of (lui) ha detto·, Valentini 1992: 150). It is possible that various reasons (not only a compensatory function) explain this pattern. As for textual functions, general discourse principles and LI influence seem to be relevant for these Chinese learners: full NPs are used for switch and shifting reference, zero subject (and later verbal agreement) for referential continuity and in cases of thematic subjects and pragmatically inferable referents (as happens in Chinese; Valentini 1992: 137-163); clitic objects for known topics are often omitted (as in Chinese, where object zero anaphora is allowed, unlike in Italian). Hendriks's (1998; this vol.) studies on reference to person and space in narrative discourse also deal with Chinese learners (Chinese LI, French L2, English L2, German L2), and compare them with native speakers. These learners mark information status from very early on, also by means of global marking (i.e. position in the clause: given information preverbal, new information postverbal), unlike little Chinese children, but to the same extent as German adults do. The only clear progression in German L2 concerns the increasing correctness of forms (gender and case marking). Local marking of newness (numerals, classifiers, articles), in L2 as in LI, is earlier and more noticeable for person than for spatial entities; reference maintaining devices are not analysed in depth by Hendriks (1998; but see Hendriks 2000). According to her, the results point to universal principles behind discourse organisation which are also valid in learners' texts, while LI influence seems to be somewhat limited or difficult to analyse in isolation from these principles. The above overview shows that factors of various kinds affect the acquisition and use of linguistic means for reference to person in L2, inter alia: textual function, topicality of the referent, L2 and LI typology (pro-drop or nonpro-drop; topic or subject-prominent), learner's morphosyntactic competence; general discourse organisational principles, iconicity and markedness. The present study aims at verifying the impact of some of these factors, dealing especially with textual functions of reference devices in the Italian L2 of German learners. We want to verify to what extent information status marking for personal referents conforms, in principle, to native speaker usage, as it seems to be in other L2 data (cf. Broeder 1991; Hendriks 1998) or whether it exhibits specific learning strategies. We also want to verify whether LI really plays no or little role in the acquisition of this domain, with only general factors (such as markedness, frequency in the input, salience of forms, universal discourse

Reference to person in learner discourse

75

principles and the like; Chaudron and Parker 1990; Munoz 1995; Hendriks 1998) being relevant.

3. Data Our corpus consists of narratives (film retellings) that were elicited after showing the learner a videotape with a short version of Charlie Chaplin's film Modern Times (about 20 minutes long). The informant watched the first half of the film with the researcher (who asked him to retell every single episode, in Italian, just after having seen it) and the second half alone; at the end, the informant recounted the second part, always in Italian, and often also made an oral summary of the whole story in German LI. The same procedure (except for the German summary) was used for collecting Italian native speaker data18. There are two groups of informants: I. Italian L2 group (learners' 9 ): 8 Erasmus students with German LI, age 22-25 (ALExia, KARen, CORnelia, FRAnz, in Pavia for 7/8 months, who were following an advanced course of Italian; ANTon, GISela, WOLfgang, CHRistine, in Pavia for 1/2 months, who were following either an intermediate or an advanced course of Italian); II. Italian LI group (natives): 13 Italian university students, three living in Pavia (= PAV: LAVinia, 20 years old; ARMando, 28 years old; ROBerta, 24 years old) and 10 in Milan (= MIL); the students from Milan retold only the second part of the film 20 . Transcriptions of the retellings were segmented into clauses (= cc.) and numbered; animated referents were analysed in respect of several syntactic, semantic and textual parameters 21 (cf. Hickmann et al. 1994; von Stutterheim and Klein 1989; more details on the methodology in Chini 1998b, 1999b). For the present study we will concentrate on the following parameters: I. Referential movement: referent introduction/first mention (= ri); maintenance of reference (= mr); shifting reference (= sr); II. NP referential forms: INDNOM = indefinite NP; NPRO = proper noun; DEFNOM = definite NP (introduced by a definite article); DEMNOM = definite NP introduced by a demonstrative; REL = relative pronoun; PR03 = 3rd person stressed pronoun; CLIT = clitic pronoun; ZEROP = zero (subject) pronoun (in a finite clause; verbal agreement codifies the subject); ZERO = zero anaphora (in a non-finite clause); POSS = possessive. In the

76

Marina Chini description of our data we will reserve the label "zero anaphora" for the empty subject of non-finite clauses (ZERO), following Berretta's (1990a: 94-95) analysis of anaphoric chains in Italian, and distinguish it from the zero subject pronoun of finite clauses (ZEROP). In reporting other researchers' studies we always respect their terminological choice which often resorts to the term zero anaphora for both cases.

In the following section we shall examine referential movement and referential forms employed for selected referents (animates and inanimates), for the protagonist and, in general, for people, made by both the whole group of informants and by individual subjects.

4. Results 4.1. Referent introduction: selected

referents

We first look at the means employed to introduce people and things. We restricted our analyses of the data to three animate referents (Charlie, the protagonist; the girl, co-protagonist; a woman who watched the girl stealing a loaf of bread) and four inanimate referents (the flag, the ship, the bananas, and the cabin). In the following Table we shall compare the referential means employed by 7 learners in their L I , German (only 7 of the 8 subjects also told the story in German), and of all 8 learners in Italian L2, with the forms employed by Italian native speakers (3 subjects from Pavia, PAV, whose elicitation conditions were absolutely identical to those of the learners; and the whole group of native speakers, PAV+MIL, totalling 13 subjects 22 ). Percentages, which especially in these first result sections are based on a low number of items, are reported here only to give an initial quantitative idea concerning general tendencies. The figures show a strong tendency towards employing lexical means for referent introduction, that is, most of all, indefinite NPs (66-81 % of the cases) and proper nouns (13-15% of the cases), only for the protagonist, Charlie (who could be assumed as given even before the movie started). The preference for indefinite NPs for the initial mention of referents conforms to general tendencies that have also been found in other learner varieties (cf. Hendriks 1998). Our learners also show some deviant definite NPs in this context, but definitely for a lower proportion than in Broeder's (1991) Dutch L2 data. On the other hand the use of definite NPs with a possessive expression are perfectly adequate for initial mentions if (as it is in our corpus) the "possessor" is

Reference to person in learner discourse

77

an already known referent (i.e. FRA, c. 177: vanno alia casa dell a ragazza '(they) go to the girl's house'). Some apparently deviant choices by some of the Italian subjects from the Milan group (last column to the right), i.e. some rare introductions of the two protagonists by means of pronouns or definite NPs, can be explained by the slightly different elicitation conditions (cf. supra): some members of this group do not properly treat the two main characters as new because they were already present in the first part of the film which these informants saw with the researcher. Table 1.

Linguistic means for referent introduction employed for various (animate and inanimate) referents; absolute numbers23 and percentages Italian L2 Ital. LI Ital. LI German LI Language Pavia + Pavia Milano 3 13 Nr. of subj. 7 8 % Ab % Ab % Ab Ab % s. s. s. s. 3 14 9 15 npro 4 15 7 13 17 81 39 66 indnom 42 22 81.5 76 — ~ ~ — 6 10 defnom 5 9 possnom 1 4 1 2 1 5 2 3 1 2 pro 3 1 2 zerop 1 2 clit 21 Total 27 55 59

These data confirm Givon's (1983) scale in its lower part: new topics attract lexically heavy means in L2 as in L I . Unlike little children, German adult learners of Italian do not have difficulties in choosing the right forms for referent introductions, overextending only some definite N P (but definite NPs with the possessive are correct for this function). Reasons for their good results are, presumably, not only their cognitive maturity, but also the fact that their LI resorts to very similar referential devices for this function. Also postverbal position is a frequent syntactic, global correlate of first mentions in Italian that does not cause problems for learners 24 , as it is also frequent in German L I ; the same result was found by Hendriks (1998) in Chinese learners of German.

78

Marina Chini

4.2. Reference maintenance:

selected

referents

We turn now to previously selected referents to gain an initial impression of the reference maintaining devices which are employed just after the introduction of the referent, that is in the following clause or in the next one, at a minimal referential distance of 1 or 2 clauses (other instances are not included in Table 2). Given that the referent is maximally accessible in such a context, light referential means ought to be favoured. Data show here a less uniform behavior in the three groups than in the case of reference introduction. Table 2.

Reference maintaining devices employed for various (animate and inanimate referents: global data in percentages. Italian 1Λ German LI Italian L2 (7 subjects) (8 subjects) (13 sub ects) % % Abs. % Abs. Abs. Lexical means — — 1 2 15 npro 3 4 8 defnom 4 21 11 26 4 1 5 5 12 2 demnom ~ — — — 2 5 indnom ~ — 1 2 possnom 1 2 Pronominal means 14 28 58 rel 10 53 33 - — 1 5 2 5 pro 3 — — — — 10 clit 5 Empty means - — zerop 8 19 6 12.5 — — — 1 2 zero Total 19 48 43

The figures are not high because not every subject maintained reference to the selected entities immediately after the first mention. Due to the limited number of tokens, we will comment briefly only on the major trends emerging from the Table. First of all, we can see an important difference between the groups: lexical means (the first five types) are prevalent in Italian L2 (about 44%) and important in German LI (42%), whereas Italian native speakers employ them with markedly less frequency (17%). So these German speakers' narrations (in L2 as in L I ) seem to be more explicit and redundant in coreferential contexts than the Italians' narrations. Is it possible that this preference is transferred from German LI to Italian L2 data? We will check this later on. A larger use of demonstratives in L2 than in LI also needs to be checked.

Reference to person in learner discourse

79

For pronominal means, some striking differences emerge as well. Learners resort much less often to relative pronouns in Italian L2 (32%) than in German LI (52%), although they are retelling the same content. Here the pattern of German LI should have pushed the learners in the same direction as the Italian pattern (in Italian LI texts we often find relative pronouns as reference maintaining devices, in almost 6 cases out of 10: 58%). But this happens (if at all) only to a limited extent. Another important difference between Italian L2 and LI concerns clitic pronouns: our learners never resort to them (here), while Italian speakers normally employ them for reference maintaining in (direct or indirect) object position (10%). Such a reduced use of clitics and of relative pronouns could be ascribed to learners' still incomplete (morpho)syntactic competence or to other factors which are still to be discovered. Empty means such as zero pronouns (that is, null subject pronouns in a finite clause, where verbal agreement codifies the subject) are used as reference maintaining in Italian L2 (18%) as well as in Italian LI (12%), while in German L1 they are never used for these referents (we find them in other contexts, but not so frequently). It might seem surprising that learners employ zero pronouns even more often than native speakers, given that their LI is a non prodrop language. A careful look at the data shows that this happens often in contexts where natives would resort to relative (or pseudorelative) pronouns, a coreferential means which seems important but not prevalent in learners' texts. We allude to utterances like the following: c 'e Ch. Chaplin e 0 (ZEROP) trova una bandiera 'there is Ch. Chaplin and 0 (he) finds a flag', produced by the learner KAR (at the clauses 1-2 (= cc. 1-2)); in such a context a native would have preferred a structure like: c e Ch. Chaplin che trova una bandiera 'there is Ch. Chaplin who finds a flag'. As for zero anaphora in non-finite subordinate clauses they are absent in Italian L2 and rare in Italian LI (in the case of these referents). But a more extensive examination of the whole range of referents would be necessary to verify this finding. To document learner and native speaker choices, we give here two parallel passages: one of a rather advanced learner, CORnelia, the second one by ARMando, a native speaker. Although we chose a rather proficient learner and one of the most informal narrations by an Italian native speaker, we can nevertheless notice some differences. CORnelia employs mostly correct, but sometimes too explicit, reference maintaining forms (cf. the use of the redundant demonstrative N P in c. 8 in (2) for the inanimate referent flag). The Italian native ARMando in general uses less explicit coreferential means (clitics, zero pronouns and zero anaphors). In passing, it can also be added that his passage

80

Marina Chini

also shows a slightly more cohesive and hierarchical structure (cf. the last two subordinated clauses; Chini 1998b, 1999a). (2)

Cor: passa eh un camiorij

(Verb-Subject inversion: ri INDNOM)

pass a lorry cliej (mr REL) ha per un segno di pericoloso (per?) that have as a sign of danger un/un grande pezzo di legno una banderola rossaj (ri INDNOM) [= standard It: bandiera/banderuola] a big piece of wood a flag red e questo momento come Of (mr ZEROP) passa and that moment 0 pass per questo uomo/per Charlie Chaplinc by that man by (sr D E M N O M / N P R O ) [0 = lui per/ il camionej he the lorry banderolej (sr DEMNOM) flag prende take

questo that

Charlie Chaplin a lorry] (mr DEFNOM)perde loose e: Charlie Chaplinc

questo that (sr NPRO)

and Charlie Chaplin banderolej (mr D E M N O M ) flag

'a lorry passes by that has a flagj as a danger signal - (for?) a big piece of wood and this moment as (it) passes by this man, by Charlie Chaplin c.

He loose/the lorry looses this flag and Charlie Chaplin c takes this

flag', (cc. 4-8) (3)

Arm: passa un

camionj

Pass a lorry Of (mr ZEROP) perde 0 una

bandieraj

a

flag

(Verb-Subject inversion: ir INDNOM) un: eh

diciamo+

loose a we say (perj segnalazione for

signal

(ir INDNOM)

Reference to person in learner discourse + Charlie Chaylinc

(sr NPRO) prende

Charlie Chaplin

take

bandieruj

ad

agitar

flag (mrDEMNOM) Idj (mr CLITOB)

begin to Ος (mr ZERO) avvertire:

move penso

it il

Og (mr ZEROP) comincia 0 per

'sta

81

that

for 0 camionista lorry driver

warn I think the (sr DEFNOM) (che = REL or COMPL)

l'j ha

persaj

(mr CLITOB; pst part.: gdr.agr. object)

it have

loose

'a lorry passes by, (it) looses a - let's say - a flagy for signalling. Charlie Chaplin c takes this flagj, (0) c begins to shake itj to 0 C inform - 1 think - the lorry-driver (that he) lost itj

(cc. 3-8)

This preliminary survey of some referents shows that there are sectors in which learners' referential choices are, or begin to be, similar to native speakers' (zero pronouns), and others for which we observe an acquisitional difficulty (clitic and relative pronouns) and/or a possible source language influence that can lead to considerable differences in referential strategies. Reference maintaining devices seem to be less native speaker-like than those employed for referent introduction.

4.3. Reference maintenance

to persons (whole

corpus)

In order to gain a larger picture of reference maintenance in the corpus, we shall now consider the means used for referents that have already been mentioned in the text, but not necessarily in the immediately preceding context. More precisely, we shall look for the presence of some of the most significant reference maintaining devices for persons in the whole corpus of retellings in Italian L2 and Italian LI (10 Milan subjects together = MIL). Here we exclude the normal definite NPs (with a coreferential function), which are about four times more frequently used by learners than by Italian natives, and concentrate on other more particular and, from our point of view, more interesting means: NPs with demonstratives and especially pronominal and empty devices, which seem to be used somewhat differently by learners and native speakers. In the following two tables learners are considered separately (Table 4) and grouped

82

Marina Chini

according to their growing morphosyntactic competence (Table 5)26. The percentages concern the presence of the various means per 100 clauses (= %cc.), that is, their frequency in the corpus; percentages are only indicative and are supplied essentially to make the comparison easier. Results confirm the tendencies commented on above which here gain a more general significance. In order to obtain a global impression we can compare the frequency of use of the various devices in Italian L2 (% in bold in Table 4) and in Italian LI (% in bold in Table 5). It transpires that, as far as reference maintenance is concerned, learners resort more than native speakers to heavier/stressed means (NPs with demonstratives and full (subject) pronouns), and less to light/unstressed or empty means (relative pronouns, clitics and zero pronouns). As for demonstratives, learners employ them 3-4 times more than Italian native speakers do. Italian native speakers use NPs with demonstratives almost exclusively when the preceding mention is more than 1 clause distant. The frequence of demonstratives would have been even higher in the learners' texts if we had considered the case of inanimate referents (4-6 cases every 100 cc. in ANT, cf. ex. 4; see also COR, ex. 2, cc. 7, 8). (4)

Ant: e: per caso un/un/una m/una macchina passa And by chance one car pass e questa macchina (,mr-DEMNOM) - perde: And that car loose una bandiere (instead of: bandiera) a flag e: Charlie Chap/Charlie vuole: eh dare una and Charlie want give a mano a: questoJquesto:[,..]uomo hand to that man 'By chance - a car passes by and this car looses a flag and Charlie wants to help this man' (cc. 5-7)

Reference maintenance through relative pronouns is clearly less frequent in learners than in native speakers. In L2 data we find mainly the relativiser che as subject (sometimes also the incorrect chi instead of che, ex. 5, c. 159) or as object (ex. 5, c. 155), other relative forms {dove, in cui, nel quale, etc.) being more rare here (but not so in Italian native speaker narratives; cf. Chini 1998b). One example from a relatively advanced learner:

Reference to person in learner discourse

83

(5) Ale:

lui non trova un He not find che (mr REL OBJ) that ma 0 vede dopo un but see after a che: (mr REL SUB) that mae:/fa parte di but make part of

Table 3.

Reference maintaining devices (mr) for persons in Italian L2 (8 subjects): presence in the whole retelling in absolute figures and in percentages (per 100 clauses = % cc.)

Subjects Means Demnom Rel Pro3 -lui Sub -lei -loro Pro. Total Clitobj -lo -la -li/le Clit -gli Indo -le Clit. Tot Zerop 3p Tot Tot claus.

pezzettino di legno: a tiny piece of wood 0 pud usare 0 can serve po' di tempo - un pezzettino little of time one small piece e giusto is right una un cons/construzione a a construction

1 CH

%

Gl

KA

4

4

1

2

4

2

2

-

19

1

24

20

10

15

31

12

4

22

138

7

51

42

17

25

38

46

20

5

295

15

7

1

1

-

-

10 7

9

-

-

25 2

10 6

3 4

66 19

3 1

58

43

18

42

47

73

36

12

380

20

2 1

-

8 2

1 1

5

2

-

-

-

1 1 1

1

-

-

-

-

1

18 7 2

1 0 0

-

1

-

-

-

-4

-

1

6

0

1

-

-

-

2

1

-

-

4

0

I

4

2

3

3

15

2

7

37

2

40

44

18

40

71

74

76

72

435

23

127

115

49

102

156

176

120

113

958

29

319

98

15

31

36

21

20

192

-

AL

Tot

WO

-

CO

FR

AN

-

84

Marina Chini

Table 4.

Reference maintaining devices (mr) for persons: presence in absolute figures and in percentages (per 100 clauses) in 3 Italian L2 groups (% in italics) and in the MIL group of Italian LI (% in bold) 2 3 (CO+AL M1LANO Groups 1 (WO+GI) (AN+KA +FR) 10 subjects +CH) Ab Means Abs. % % Abs. % Abs % s. 1 1 Demnom 5 8 6 0.7 3 0.3 Rel 30 7 43 6 65 8 150 13 Pro3 -lui 14 96 14 89 59 10 46 4 Sub -lei 2 elopment, 225-240. Cambridge: Cambridge University Press. Marslen-Wilson, William, Elena Levy and Lorraine Komisarjevsky Tyler 1982 Producing interpretable discourse: the establishment and maintenance of reference. In: Robert Jarvella and Wolfgang Klein (Eds.), Speech, Place and Action. Studies in Deixis and Related Topics, 339-378. New York: Wiley. Meisel, Jürgen 1991 Principles of Universal Grammar and strategies of language use. On some similarities and differences between first and second language acquisition. In: Lynn Eubank (ed.), Point Counterpoint. Universal Grammar in the Second Language, 231-276. Amsterdam: Benjamins.

108

Marina Chini

Min, Ruifang 1994 The acquisition of referring expressions by young Chinese children. Unpublished doctoral thesis, Nijmegen. Munoz, Carmen 1995 Markedness and the acquisition of referential forms: The case of zero anaphora. Studies in Second Language Acquisition 17: 517-527. Munoz, Carmen 2000 The over-explicitation of personal reference in L2 narratives. Paper presented at the Euroconference on Information structure, Linguistic Structure and the Dynamics of Acquisition, San Feliu de Guixols, october 2000. Orsolini, Margherita and Paola Di Giacinto 1996 Use of referential expressions in 4-year-old children's narratives: invented versus recalled stories. In: Clotilde Pontecorvo, Margherita Orsolini, Barbara Bürge and Lauren Resnik (eds.), Children's Early Text Construction, 67-81. Mahwah NJ: Erlbaum. Orsolini, Margherita, Franca Rossi and Clotilde Pontecorvo 1996 Re-introduction of referents in Italian children's narratives. Journal of Child Language 23: 465-486. Perdue, Clive (ed.) 1993 Adult Language Acquisition: Cross-linguistic Perspectives. 2 voll. Cambridge: Cambridge University Press. Pfaff, Carol 1987 Functional approaches to interlanguage. In: Carroll Pfaff (ed.), First and Second Language Acquisition Processes, 81-102. Cambridge MA: Newbury House. Phinney, Marianne 1987 The pro-drop parameter in second language acquisition. In: Thomas Roeper and Edwin Williams (ed.), Parameter Setting, 221-238. Dordrecht: Reidel. Pizzuto, Elena and Maria Christina Caselli 1992 The acquisition of Italian morphology: Implications for models of language development. Journal of Child Language 19: 491-557. Polio, Charlene 1995 Acquiring nothing? The use of zero pronouns by normative speakers of Chinese and the implications for the acquisition of nominal reference. Studies in Second Language Acquisition 17: 353-377. Power, Richard and M.F. Dal Martello 1986 The use of the definite and indefinite articles by Italian preschool children. Journal of Child Language 13: 145-154. Renzi, Lorenzo (ed.) 1988 Grande grammatica italiana di consultazione. Volume I: Lafrase. Isintagmi nominale e preposizionale. Bologna: II Mulino.

Reference to person in learner discourse

109

Sterelny, Kim 1994 Reference, philosophical issues concerning. In: R.E. Asher and J.M.Y. Simpson (eds.), The Encyclopedia of Language and Linguistics. Vol. 7, 3492-3500. Oxford: Pergamon Press. von Stutterheim, Christiane 1997 Einige Prinzipien des Textaufbaus. Tübingen: Niemeyer. von Stutterheim, Christiane and Wolfgang Klein 1989 Referential movement in descriptive and narrative discourse. In: Rainer Dietrich and Carl Graumann (Eds.), Language Processing in Social Context, 39-76. Amsterdam: North-Holland. Tomlin, Russell (ed.) 1987 Coherence and Grounding in Discourse. Amsterdam: Benjamins. Tsimpli, Ianthi and Roussou, Anna 1991 Parameter-resetting in L2? UCL Working Papers in Linguistics 3: 149169. Valentini, Ada 1992 L'italiano dei cinesi. Questioni di sintassi. Milano: Guerini. Valentini, Ada 1994 Soggetti pronominali nell'italiano L2 di sinofoni. In: Anna Giacalone Ramat and Massimo Vedovelli (eds.), Italiano: lingua seconda, lingua straniera, 297-318. Roma: Bulzoni. Van Valin, Robert 1987 Aspects of the interaction of syntax and pragmatics: discourse coreference mechanisms and the typology of grammatical systems. In: Jef Verschueren and Marcella Bertuccelli-Papi (eds.), The Pragmatic Perspective. Selected Papers from the 1985 International Pragmatics Conference, 513-531. Amsterdam: Benjamins. Veronique, Daniel 1989/90 L'apprentissage du franfais par des travailleurs arabophones et la genese des Creoles "franfais". Langage etSociete 50-51: 9-37. White, Lydia 1985 The 'pro-drop' parameter in adult second language acquisition. Language Learning 35: 47-62. Wigglesworth, Gillian 1990 Children's narrative acquisition: A study of some aspects of reference and anaphora. First Language 10: 105-125. Williams, Jessica 1988 Zero anaphora in second language acquisition. A comparison among three varieties of English. Studies in Second Language Acquisition 10: 339-370.

110

Marina Chini

Appendix List of abbreviations c./cc. clause/clauses CLIT clitic pronoun DEFNOM definite NP (introduced by a definite article) DEMNOM definite NP introduced by a demonstrative INDNOM indefinite NP mr maintenance of reference NPRO proper noun OBJ object POSS possessive PR03 3rd person stressed pronoun REL relative pronoun ri referent introduction/first mention sr shifting reference SUB subject ZERO zero anaphora (in a non-finite clause) ZEROP zero (subject) pronoun (in a finite clause; verbal agreement codifies the subject) ιRotational conventions used in the examples suspensive intonation + ++ +++ pauses of growing length (to 3 seconds) / self-correction or self-interruption by the speaker () less audible item(s) [] remark of the transcriptor [...] expunction eh, ehm hesitation :, :: lengthening

Structuring space in discourse: A comparison of Chinese, English, French and German LI and English, French and German L2 acquisition1. Henriette Hendriks 1. Reference to space and the narrative task This study analyses the way in which adults learning a second language and children learning their first language acquire how to spatially relate events in a narrative 2 . Creating a functioning spatial surrounding for events involves a number of capacities such as 1) cognitively representing the various spatial constellations; 2) finding the appropriate linguistic means to express those spatial constellations in the language being acquired; and 3) keeping track of the spatial information across utterances in order for the narrative to be coherent with respect to the spatial domain. One of the more basic definitions of a narrative states that a narrative consists of a sequence of temporally ordered events. Given this definition, there does not seem to be much room for the analysis of space in a narrative, as space is apparently not one of the main organizing domains. A simple example such as (1) will make it clear, however, that first of all, although the features of the example correspond to the definition, it is far from what we normally expect a narrative to be, and that secondly, even in this minimal form, the text includes spatial information concerning the protagonists' moving from one to another location (from the here, with us, to the not here, i.e., their home). (1) Alison and Alex came over. We had brunch and then they went again.

home

A number of researchers have therefore tried to formulate a more comprehensive definition of narrative. One line of research (Klein and von Stutterheim 1987) proposes that the structure of all coherent texts is constrained on both global and local levels by the nature of the question — the Quaestio — which the text in its entirety is meant to answer. According to this approach, any utterance in the text integrates information from a combination of domains, the particular Quaestio influencing the way in which possible domains of reference are realised at utterance and text level, and the development of the domains across utterances. For a narrative, the temporal domain will be basic for the organisation of the text, three other domains also being crucially in-

112

Henriette Hendriks

volved, however, namely: space,persons (and objects), and events. For a route description, the spatial domain will be more basic. Most approaches have in common the idea that some clauses in the text carry foregrounded information, others backgrounded information. The Quaestio approach, for example, assumes that utterances directly answering the Quaestio (in a narrative those pushing the time line forward by commenting on the sequence of events) form the foreground. Although we do not feel that the foreground / background distinction is an easy one to make, it is important to know here that spatial information in a narrative may end up being considered backgrounded, which probably explains why narratives can be perfectly understandable even when lacking a spatial setting, as can be seen in example (2) below (note, however, that although no explicit locations are mentioned, motion events are referred to). (2)

Um this is the bird... with chicks. And he's flying away to get the chicks something to eat. And this is the chicks and the cat... and the dog right over here. And the cat trying to climb to try to get them. Then the bird flies back... with a worm ... the cat - and then the cat and the dog (Ad: uh hum) and then the dog... and then the birds chased the dog and the cat away (English 4-year-old).

When referring to space in a narrative, its main function is to localise protagonists, entities, and the events in which they are involved. Given that we are dealing with texts rather than with single utterances, the spatial information may be provided across utterances and be understood depending on the construction of the discourse and our understanding of the discoursepragmatic principles guiding this construction. These principles require that, if possible, a spatial surrounding should be provided. If we follow the idea of a story grammar (Mandler and Johnson 1977; Stein and Glenn 1979; Stein 1982), we might even expect this spatial information to turn up in the socalled setting, before the story really starts developing. Furthermore, in a situation in which speaker and hearer do not share any mutual knowledge about the contents of the narrative, discourse pragmatic principles require that all new information be marked as such. In the European languages that we are about to examine (English, French and German) such information is expected to be marked locally with nominals with indefinite articles. In Chinese, the 4th language in this study, it is marked globally with postverbal position. Once the spatial frame is set, it may be inferred and need not be explicitly mentioned anymore. Only changes of location involving a change of spatial frame must be marked more or less explicitly, depending on discourse needs. Therefore,

Structuring space in discourse

113

frequent utterances without any explicit reference to spatial locations are expected, as long as the events reported take place within the boundaries of a non-changing space. When locations are explicitly referred to, it can be with less explicit forms, e.g., pronominal rather than nominal ones. How to distribute the spatial information in order for it to be available when necessary but not too explicit when superfluous is thus related to acquiring the discoursepragmatic principles in general. Although these general discourse rules apply to all languages, several other factors are also likely to have an impact on reference maintenance. First, languages differ in their tolerance of more or less explicit forms. Thus, Chinese is more discourse-oriented than the European languages concerned here, one of the results being that it sets fewer restrictions on the use of zero forms. Languages also organise the packaging of spatial information in importantly different ways, and it has been shown that this will also influence the level of explicitness in reference maintenance, as we will see in the next section. Finally, the contents of the story obviously will also have an impact on the spatial organisation. Thus, in the present CAT story (cf. Appendix I), if we assume the tree to be the most general possible location for protagonists and events to take place, then the mother bird and baby birds are actually in that tree, but the cat and dog are under, on, around, or next to the tree. Moreover, the mother and babies are in the spatial frame right from the beginning of the story. The cat and dog, however, come into the spatial frame only later, thereby necessarily changing location from not being at the scene to being at the scene. All this variation in spatial constellations will influence the choice of linguistic means at utterance level and the organisation of spatial information across utterances.

2. Cross-linguistic variation in reference to space Languages have a large number of linguistic means to express how one entity (the Figure) is located with respect to some other entity (the Ground) (terminology as introduced by Talmy 1975, 1983, 1985). Although these linguistic means reflect the conceptualisation of space by the human mind to some extent, they leave enough room for language-specific distinctions in the encoding of spatial information (Behrens 2001). To mention just some differences, some languages encode all information in the verb root, others allow serial verb constructions, and yet others supply a whole array of prepositions, verbparticles, etc.

114

Henriette Hendriks

Some of this variation has been discussed in more detail and length than other types of variation. Thus, since Talmy introduced his distinction between verb-framed and satellite-framed languages (Talmy 1991), arguing that Romance languages are an example of the former and Germanic languages and Chinese are an example of the latter, a number of studies have tried to explicate the effects of this distinction on the organisation of spatial information in general and more specifically also on the discourse organisation in these types of languages. The distinction according to Talmy is concerned with the properties described below. Motion (or locating) situations express essential information such Motion or non-motion (location) itself, and the Path of motion, and, less central to the motion event, the Manner and Cause of motion. This conglomerate of spatially related information has to be "packaged" in a language-specific way when space is referred to. Verb-framed languages more characteristically encode the more fundamental parts of spatial information (Motion and Path) in the verb itself, whereas satellite-framed languages rather encode the Path in the satellite. A few words should be said here about what Talmy considers to be satellites. It includes elements familiar to European languages such as prepositions, adverbs and verb-particles, but in Chinese, it also includes parts of the so-called Resultative Verb Construction (RVC). The RVC is a sequence of two (or three) nonidentical verbs immediately following each other, i.e.: Verb 1 -Verb2(-Verb3) All these verbs can occur alone as main verbs, but can also occur in an RVC. In the RVC, the first verb can be basically any kind of verb (encoding cause, manner, path, etc.). The second verb, the so-called verb-complement and element considered a satellite by Talmy, can express a path or reaching of the spatial goal. When a verb of locomotion occurs in the first position, it is readily followed by directional complements (satellites) as exemplified in (3). (3)

Mao ρά-shang shii Cat crawl-ascend tree 'The cat climbs up the tree'

As a result of the typological verb/satellite difference, the remaining information associated with motion, such as Manner and Cause, ends up in other parts of speech. In satellite-framed languages, Manner is frequently encoded in the verb root, resulting in a large number of verbs expressing a variety of Man-

Structuring space in discourse

115

ners, such as English jump, crawl, hop, and Chinese pa ('crawl') as in (3) above. In verb-framed languages, Manner and Cause tend to either not be expressed at all (it being less essential information), or to be expressed in a separate clause or gerundive type construction as in: en sautant, en courant, en nageant, etc. Talmy intended to speak of "tendencies" as far as packaging is concerned, and since his work was published, a number of researchers such as Aske, Slobin and Hoiting, and Naigles et al. have tried to find reasons for the tendencies to be followed or not. Results of these further studies show that the Romance languages have a system in which some motion verbs conflate Manner and some conflate Path (Aske 1989), the ones conflating Manner being restricted in use. More specifically, Slobin and Hoiting (1994) claim that Path conflating verbs are the obligatory option for all events that involve "entering", "exiting" or "crossing a boundary" (cf. also Hickmann 2003). Naigles et al (1998), finally, show how these differences in information encoded in the verb root change utterance constructions referring to space in general. The satelliteframed languages typically provide more elaborate descriptions, the Manner verbs in those languages typically being accompanied with satellites appearing with or without explicit nouns referring to the Ground. In contrast, the Romance languages typically show bare verbs both when encoding Path and when encoding Manner. Some researchers, such as Slobin (1991), claim that differences in linguistic means at predicate and/or utterance level, will also have an influence on the discourse organisational level. As he puts it: "In the time frame of constructing utterances in discourse, one fits one's thoughts into available linguistic frames. 'Thinking for speaking' involves picking the characteristics of objects and events that a. fit some conceptualisation of the event, and b. are readily encodable in the language". It was found that, when confronted with conceptually complex events such as in examples (4) and (5) below, speakers of satellite framed languages tend to encode the situation using dense packaging of information in the verb root and a conglomerate of satellites, these means being readily available in their language (as in example (4)), whereas Romance language speaking natives opt for a distribution of information, creating a series of partly static descriptive utterances, partly a single verb-root expressing the vertical path, as in (5). (4)

The boy was thrown down the cliff into the river below by the deer.

116

Henriette Hendriks

(5)

Le gargon se trouve en haut de la colline. The boy refl. find on top of the hill En has il y a line riviere. Below there's a river Le cerf le pousse, et il tombe. The deer him pushes, and he falls 'The boy finds himself on the top of the hill. Below there is a river. The deer pushes him, and he falls'.

Carroll and von Stutterheim (1993) show that, when analyzing reference to space in two space-typologically very similar languages, English and German (both satellite-framed), the texts constructed for the same linguistic task in those two languages are also strikingly different. In English the object domain, that is, the objects and their intrinsic features, control the spatial domain, whereas in German the deictically structured space has more autonomy, that is, it can, in the process of text building, provide the basis, the background against which objects can be placed. These differences cannot possibly be a result of the availability of linguistic means in one language vs. the other, since both languages have means to relate entities intrinsically and/or deictically. Given the variation in reference to space in the different languages, we assume that learners will be confronted with different learning problems when tackling the acquisition task. In the following we will try and formulate some more specific acquisition problems and hypothesise how they may be solved by the different groups of learners.

3. The acquisition task and some hypotheses We started this paper by saying that, in order to structure space in discourse, speakers need to acquire a number of capacities. The first of those concerned the capacity to form a cognitive representation of the spatial constellations involved in the narrative. A large number of studies concerning this capacity in the child exist, a most influential one by Piaget and Inhelder (1947). Findings of such studies show that well before children start speaking, they know a great deal about space. With the help of non-linguistic tests, Piaget and Inhelder established the emergence of spatial concepts such as containment, support, proximity, etc., well before the age of four. This does not mean, however, that the child's understanding of space at four years is adult-like. It is

Structuring space in discourse

117

only when it reaches the level of representational thought, which is free of space and time, that the child can be said to operate in a world in which it sees itself as just one point in a bigger matrix full of inter-related points (as do adults), rather than as the point from which all relations depart. For the purposes of the current study, however, the children seem conceptually efficiently equipped. If we assume that the cognitive representation of spatial constellations is universal and not language specific (as Carroll and von Stutterheim may want to suggest), adults learning a second language do not have to acquire this capacity anymore, because once acquired it serves for the rest of their lives. The second capacity we mentioned concerned the acquisition of appropriate forms to express the spatial constellations. Again, a large number of studies have analysed the child's acquisition of those forms, that is, prepositions, verbs, verb-particles, etc. Cross-linguistic studies (Slobin 1973; Johnston and Slobin 1979) compared the acquisition of spatial expressions in up to nineteen different languages, and found that children at very similar ages acquire forms for the same type of spatial constellations, starting cross-linguistically with forms to express the containment concept, then moving on to the support concept, vicinity, and the front-back axis. The latter pair of spatial expressions is acquired around 4;8. A large number of studies has investigated the driving forces behind the acquisition of spatial terms. In the eighties, based on the cross-linguistic regularities found in the order of acquisition of similar spatial terms, it was believed that the concept was acquired first, triggering the child's search for a linguistic term that would fit that particular concept. Recently, however, researchers like Bowerman and Choi (2001) have pointed out that crosslinguistic variation is much more important than believed in the eighties, some languages not having a linguistic means to express the presumedly "basic" concepts like containment. These facts make it highly unlikely that underlying concepts alone drive the acquisition of spatial expressions. Irrespective of the driving force behind the acquisition of the linguistic forms, children by the age of four have usually acquired a minimal set of the spatial expressions available in their language. Consequently, the acquisition of forms should not be a problem for our LI learners either. Note that even though the studies I have just discussed focus on the acquisition of prepositions, further studies by Slobin (1991) show that the acquisition of the verbal variation and the resulting differences in utterance and discourse organisation, are acquired by children from a very young age on in language-specific ways, such that their linguistic productions correspond remarkably well to the target language input.

118

Henriette Hendriks

Where children have thus been claimed to be highly sensitive to the target language input, and to structure space from early on as their mother tongue requires, we propose that for adult second language learners this capacity will pose problems. Languages being as varied as they are, it seems obvious that adults will need time to adjust to the target language system, especially if the system differs significantly from the source language. If we have a look at those differences, we find that Chinese learners of German or English will find atypologically related language as far as the expression of space is concerned, in that all three languages, according to Talmy, are satellite-framed. In contrast, Chinese learners of French will have to adjust to the typologically different French verb-framed system. In concreto, these differences may involve the following: Speakers of a satellite-framed language might be able to provide more spatial information (through extensive packaging) whereas speakers of a verb-framed language might have to make choices about what information to give, or divide information over several clauses in discourse. On the one hand, Chinese learning French may thus occasionally want to cluster more information than habitually found in the French native speaker. On the other hand, satellite-framed languages, producing more complex spatial constellations on the utterance level (compare (4) and (5) above), may present the learner with an overwhelming task of having to find a way to match the variety of linguistic means with the corresponding concepts. Moreover, satellites coming in rather different types across languages, this may provide an additional acquisitional problem. Finally, the third capacity concerned the acquisition of discourse-pragmatic principles to be used to introduce and maintain reference to space in the narrative (among others). Previous studies in this domain concentrating on reference to person have clearly shown (Hickmann 1995; Hickmann et al. 1996, Hickmann and Hendriks 1999) that these principles are a relatively late acquisition in the child (after 7 years of age). Although children are not entirely ignorant about the knowledge shared (or not) between listener and hearer and other discourse pragmatic principles, it takes them a long time to acquire the pluri-functionality of linguistic forms such that they can use them not only on the utterance level but also to appropriately mark discourse-pragmatic principles. Studies concerning the introduction of spatial information in discourse (Hendriks 1993; Hickmann, Roland, and Hendriks 1998) have shown that development in LI is even later (as compared to form). Comparing the acquisition of discourse-pragmatic principles of LI and L2 learners, starting with the hypothesis that L2 learners have already acquired this capacity, the results show that adult learners do not follow the same acquisitional path as the children. As far as the distinction between new and

Structuring space in discourse

119

given information is concerned, they tend to make such a distinction from the lowest level of proficiency onward. However, concerning reference maintenance, these adult learners do not always react as expected, and use more explicit forms for maintenance than the native adult speakers in some languages (Hendriks and Hickmann 1998; Hendriks 1998; Hendriks 2002). In this study, we are dealing with Chinese native speakers learning European languages. As mentioned previously, Chinese is a discourse-oriented language, which entails a higher proportion of zero forms than found in subject-oriented languages. Thus, although the universal rule less known referent - > more explicit form better known referent - > less explicit form is valid in all four languages, language-specific differences interfere here, and the level of explicitness is bound to differ considerably between Chinese on the one hand and the European languages on the other. This has been shown true for reference to person. In this paper we will investigate if similar patterns occur with reference to space, that is, in reference to Grounds. Given previous studies, we start out with the following hypotheses: Cross-linguistic matters: In the native adult data we expect to find 1. a sufficiently elaborate amount of spatial information, structured in such a way that it also provides a spatial setting, and is marked as new information. 2. An organisation which will be language-specific to the extent that it can be explained by the variation of linguistic means available for reference to space in the four target languages. 3. A level of explicitness sufficient for good understanding, the level varying from one language to the other given the available linguistic means. Age related matters: 1. We expect children and adults both to have matured as far as spatial conceptualisation is concerned. 2. We expect children not to have any problems with the mapping of language-specific means onto spatial concepts. However, given the crosslinguistic differences discussed above, and the possible influence of the LI on the acquisition of the L2, we have to test at least the assumption that

120

Henriette Hendriks

adult L2 learners will have problems. 3. With respect to the third capacity, we expect to find development in the L1 data, but not in the L2 data, except maybe for reference maintenance (where cross-linguistic differences might influence the acquisition process in the L2). 4. Method and Data The materials used to elicit the narratives consist of two picture sequences, a CAT and a HORSE story, of which we will only use the CAT story data in this particular paper (see appendix I). In order to ensure the absence of mutual knowledge, learners were confronted with the stories for the first time in the presence of a naive and unfamiliar interlocutor. In the case of the child data, two adult experimenters were present. One of them served as the naive interlocutor, while the other gave the instructions. The child was first asked to blindfold the interlocutor and to make sure she could not see. This procedure was meant to discourage the child from relying on deictically available information. When the child had blindfolded the interlocutor, the other adult presented him with the first story. That adult then moved away from the scene and let the child tell the story, without interfering with the experiment anymore. The same procedure was repeated for the second story. The order of presentation of those stories was counterbalanced. Adult subjects were told that a naive interlocutor would listen to a tape-recorded version of their narration, and had to be able to understand it without having access to the pictures. The following subject groups were involved: 1. Monolingual Chinese, French, English and German adult native speakers; 2. Monolingual Chinese, English, French and German children in three age groups (10 children per age group); 3. Chinese adults acquiring French (20 subjects), English (20 subjects), or German (40 subjects at four different levels of proficiency). The monolingual data come from a larger database collected by M. Hickmann, as discussed in Hickmann (1995), and (2003) 3 .

5. Results 5.1. The introduction

of potential

Grounds

In the following we will briefly sketch the mechanisms used by child and adult learners for the introduction of potential spatial information in the narrative. As mentioned before, a number of previous publications concerning LI and L2 acquisition have dealt with this subject 4 and we will therefore limit our-

Structuring space in discourse

121

selves to a summary of the findings and a brief discussion. The English L2 data discussed here have not been published before. The analyses were all concerned with information potentially or actually allowing protagonists, other entities and events of the CAT story to be located. In principle, any entity can function as a Ground for a Figure. Thus, in the following example, the dog is located with respect to the cat. (6) While the cat was climbing up the tree, the dog came up behind him. However, two elements in the CAT story are particularly likely to take up the role of the Ground, the tree and the nest. Given these two potential Grounds, this first set of analyses concentrated on those two entities. However, any other entity (animate or inanimate) that was first mentioned in a locative role was included in the analyses concerning the introduction of (potential) Grounds as well. These additional entities may consist of "subparts", "sublocations" of the already introduced items, such as branches of the tree, as in example (7). It may also contain entirely new spaces (8), or some space differentiated but close to the original location, as in example (9). (7) (8)

(9)

But unfortunately she didn 7 get ...to the first branch ...(el 008cat.cod) Im Walde ... ist — eine Familie (gl007cat.cod) in the forest is a family 'In the forest... is a family' et par terre y'a l' herbe. (f0509cat.cod) and on ground there is the grass 'And on the ground there is grass'.

A first analysis concerned the amount of spatial information introduced in the narratives. As far as the LI data are concerned, it was found that in all four languages narratives tend to get more spatially elaborate with age, until adult age, when narratives always include at least one potential Ground. This tendency is clearest in French, where adult stories always include two or more possible Grounds. German adults seem least concerned with giving spatial information, in that only 60% of their narratives have two or more potential Grounds, 40% occurring with only one potential Ground. As far as the adult L2 data are concerned, the hypothesis is that these learners introduce this type of information from the first level of proficiency onward, given their previous experience with discourse organisation. Note furthermore that the Chinese native data suggest that speakers in this language provide rather elaborate spatial information from 7 years onwards. If the

122

Henriette

Hendriks

source language influences the L2, as we think likely in this aspect of the narrative task, then we should expect rather elaborate spatial information overall. The data seems to suggest that elaborateness is influenced by the target language rather than by the source language. Elaborateness is highest in French as an L2, and lowest in German as an L2, thereby corresponding to the French and German, rather than to the Chinese native data. Data concerning the development in the L2 contradict our expectations of finding elabore spatial information from the lowest level of proficiency onward. Thus, the German L2 data do show development, in that narratives at the lowest proficiency level sometimes do not provide any spatial locations, and in that only at the highest level the learners of German as an L2 provide more elaborate spatial information. Analyses concerning the appropriate marking of newness on potential Grounds in previous studies were restricted to what Hickmann (1995) calls the "local" marking of newness, that is, marking on the noun phrase (a noun with an indefinite article in the European languages; nouns preceded by a numeral and classifier in Chinese). In the LI data there is a tendency to mark this type of information appropriately more frequently in the adult than in the child data, even though the developmental path is not regular and dissimilar across languages. German adults seem less concerned with marking newness than adults in the other three languages (cf. Hickmann, Roland, and Hendriks 1998; Hendriks and Hickmann 1998). In the L2 data, given that marking newness is assumed to be a more or less universal discourse pragmatic principle, we again expected to find little development over proficiency levels, and a generally high proportion of appropriately marked entities. Data confirmed the lack of development with proficiency, but, the expectancy of high overall appropriateness of newness marking was disconfirmed in that in both English and German as an L2 it is less frequent than in the native speaker adult data. N o clear influence of either source or target language is noticeable. A final analysis asks how early the introduction of the spatial information is provided in the narrative and particularly whether it occurred in the setting of the narrative. Do narrators introduce spatial information in the so-called setting, as story-grammars would suggest is appropriate, or do they provide spatial information only when it cannot be avoided later in the narrative, i.e., the tree being mentioned only when the cat wants to climb it? Providing a spatial setting is an indication for the acquisition of discourse organisational principles. A developmental progression was found in the mention of potential Grounds early in the narrative. Thus, native adults provide information in the

Structuring space in discourse

123

setting in 75% to 90% of the cases (German adults 75%; Chinese 78%; French 83% and English 88%). Native 7-year-olds in all languages are the worst providers of a spatial setting. Adult L2 learners clearly prefer to introduce spatial information early in the story, as expected, given that it is part of the discourse pragmatic principles. Proportions are lower than for the native speakers, however, with Chinese learning English providing 59% of the information early, Chinese learning French 68% and Chinese learning German between 70% and 95%. Where German native adults seem least concerned with the construction of a spatial setting, Chinese L2 learners of German seem more preoccupied with this than Chinese learning French or English. Again, little development over proficiency was found in the L2 data. The conclusion of the previous range of analyses on newness marking has to be that entities serving as Grounds, or introduced to go on serving as Grounds later on in the narrative are not consequently marked for newness by local means. Possible explanations of this finding include the position in the story where the entity is introduced (setting vs. late). Thus, when the introduction takes place in the setting, the entity will more likely be appropriately marked for newness, when introduced later, it is less likely to be marked (cf. Hendriks 1993; Hendriks and Hickmann 1998). Another explanation concerns the semantic role of the entities upon introduction. The entities serving as Grounds, the tree, the nest, and other inanimate entities are introduced mainly in three semantic roles. They can be first mentioned as part of an existential construction, as in (10), in an object role, as in (11), or in a locative role, as in (12). It was shown that, when introduced in the first two types of roles, they are more likely to be appropriately marked for newness, while when introduced in a locative role, they are less likely to be appropriately marked. Both late introductions and introductions in locative role being relatively frequent, the patterns for the introduction of spatial entities are explained. (10) (11) (12)

There is a tree The cat looks up and sees a nest There is a mother bird sitting on the branch of a tree

Finally, when compared with the introduction of animate entities in discourse, newness marking is less consistent in all groups (monolingual and L2) and at all ages (4, 7, 10 and adults). Given that spatial information may be more backgrounded in the overall organisation of a narrative, this could be an explanation for this phenomenon.

124

Henriette Hendriks

5.2. Maintaining reference to space We now turn to reference maintenance, with particular attention to the following aspects. On the one hand, we will consider the acquisition by children and adults of the language-specific means necessary to encode the spatial events taking place in the narrative. On the other hand, we will closely examine the explicitness of Grounds in reference maintenance. As regards the first point, we will pay particular attention to two types of complexities found in the languages that may pose problems for the learners. The first type of complexity concerns the amount of packaged information that can be found in a single utterance produced by native speakers of the various languages, and by the acquirers of those languages. Secondly, we mentioned the diversity in types of satellites and clustering of satellites in the languages and the acquisition complexity this may involve. As Naigles et al. showed, these complexities are interrelated in that when the packaging is simple (as in French), the predicate tends to also be simple. In contrast, when packaging is more complex (as in English), frequently the predicate is also more complex, involving clustering of satellites. Both types of complexities, we feel, involve the second capacity (finding the appropriate linguistic means). As mentioned previously, children at the age of 4 should have acquired at least a minimum of those linguistic means, whereas they may pose problems for the lower proficiency adult learners. The analyses concerning the explicitness of the Ground, with which we will start in this section belong to the third type of capacity (keeping track of spatial information for discourse cohesion reasons) and might therefore pose problems for children rather than for adult L2 learners. In examining the explicitness of Grounds, we will analyze not only the forms themselves, but also the predicate and location type in which they occur, in order to find what factors influence this part of the structuring of space. Is it learner type, language type, predicate type, or maybe a combination of all those factors? Once a minimum of spatial information is provided, maintaining reference to a Ground can be more or less explicit. The different degrees of explicitness reflect degrees of givenness, i.e., lean forms and even complete omission are possible when reference is clear to the listener, while fuller forms are necessary otherwise. Choosing the right level of explicitness is a delicate matter. According to the previously mentioned general discourse rules, the absence of an explicit location after spatial information has been provided should lead the listener to infer there is no change of location. However, it may be the case that a new event takes place in a specific sub-location of the location already introduced. In this case, the explicit specification of this Ground may be useful

Structuring space in discourse

125

after all. In addition, choosing the right level of explicitness interacts with other factors, such as whether all events take place in one general location or whether protagonists change location. Does the change of location take place within the boundaries of the currently introduced spatial constellation or is it functioning as the goal or source of a spatial change? In the latter case, the utterance referring to this spatial scene may very well involve deictic predicates. All these factors therefore influence the explicitness of the forms. Forms were grouped into three categories, ranging from more explicit to less explicit ones: 1. Full forms: these forms consist of nominals with lexical information. In the European languages, these are nominals with definite articles. In Chinese, full forms consist of bare nominals or nominals preceded by a demonstrative determiner plus classifier. 2. Lean forms: in the European languages these forms consist of the deictic expressions here and there (hier, da, ici, la), whether or not accompanied by a preposition (darunter). Included in this group are also nominalised place words (inside), adverbials (en has), etc. In all these cases, forms function as anaphoras for a Ground already mentioned in the narrative. They all depend on context for completion of their meaning. In Chinese, lean forms consist of nominalised place words such as liton and of the deictic words zher and nar. Note that with deictic expressions, the exact spatial relation (e.g., in, on, over) is not indicated whereas such a relation is necessarily expressed in the other lean forms, such as the nominalised place word. Cf. the difference between examples (14) and (15). 3. Zero forms or omissions. A third level of explicitness involves the complete omission of the Ground, as shown in (16) and (17) for French and German respectively. (14)

In the middle of the room there was a table and a cat sat there.

(15)

Zäi fängzili you zhuözi Be-at room inside there-be table 'There is a table in the room' Yi-zhJ mäo zäishängtou One-cl cat be-at top-side 'There's one cat on top of it'

(16)

La mere 'The mother

s'envole flies away'

126

Henriette Hendriks

(17)

Und die Katze klettert 'And the cat climbs

hoch up'

5.2.1. Predicate and location types in utterances expressing motion and location 5 In a first analysis, we make an inventory of types of predicates, i.e., dynamic vs. static, as combined with type of location, i.e., general vs. change of location. This analysis is of importance as we will assess in further analyses how predicate and location type influence the explicitness of Grounds. To a large extent, the predicates chosen will be dictated by the story contents as given in the picture sequence. For example, when the mother bird disappears from the scene, the narrator is almost obliged to make mention of that departure, for story-line reasons, but also for coherence reasons, given that the mother bird re-appears in a later picture. Another factor influencing the choice of predicate and location type are the language-specific means readily available for reference to space in the particular language spoken, and the variation in packaging devices as discussed by Talmy. Keeping these preliminary remarks in mind, let us now look at Figure 1. It provides an overview of predicate / location combinations as found in the monolingual data. The figure shows that native speaker adults most frequently used motion predicates referring to a change of location (e.g., the dog runs away). Least frequently used predicates are motion predicates referring to a general location (the cat was roaming around), even though in French and English adult samples they gain importance and become more frequent than static general locations. The child data show quite some development in this area. At 4 years, static predicates referring to general locations (e.g., the babies are in the nest) are the preferred choice for maintaining reference to space in the three European languages, this type of predicate staying highly frequent in French even up to 10 years old. Note that these static predicates in the child language are mainly of an existential type (There is a dog), i.e., they acknowledge the existence of an entity in the pre-existing space, this space having been more or less unequivocally introduced in the previous discourse. Chinese children do not show this European developmental path. In fact, hardly any development is found in the choice of predicate and location type in Chinese, children at 4 years and older showing the adult preference for the use of dynamic predicates referring to a change of location (cf. also Hickmann, Roland, and Hendriks 1998 and Hendriks and Hickmann 1998 for similar results concerning the complete database (Horse and Cat))

Structuring

space in discourse

English

French

I

I

]

I

4 years

7 years

10 years

Adults

mot/chg

127

*

I 4 years

sta/gal

mot/gal

•

mot/chg

·

mot/gal

1 7 years *

1 10 years

1 Adults

sta/gal

Chinese

German

80

60

40

20-

ι

ι

4 years

7 years

1 10 years

1 Adults

4 years

7 years

mot/chg

mot/chg

— *

mot/gal

mot/gal

Figure 1. Overall distribution of situation type.

10 years sta/gal

Adults

128

Henriette Hendriks

Data of the adult Chinese L2 learners shown in Figure 2a show that the distribution of predicate / location types reflects the target language rather than the source language data. That is, whereas Chinese natives use very few motion predicates expressing general location overall (6%), Chinese learners tend to use this type of predicate slightly more frequently in the various L2 interlanguages (English (15%), French (18%), and German (14%)). 100

-

80

-

chineng Η

mot/chg

chinfren jiiiij

mot/gal

chinger [0]

sta/gal

Figure 2a. Overall distribution of situation type. L2 learners. Development over proficiency levels as attested in the German L2 data (Figure 2b) shows that adult learners, from the lowest proficiency level onward, chose their predicates "as adults" thereby reflecting both source and target language patterns. They do not go through a stage in which they merely establish existence of entities, which provides the under-specified spatial constellations in the child data.

5.2.2. Explicitness of Grounds In the next analysis, we examine the explicitness of spatial information, taking into account the factors age, language, predicate and location types.

Structuring space in discourse

level I •

mot/chg

•

mot/gal

Figure 2b.

level II A

level III

129

level IV

sta/gal

Overall distribution of situation type as a function of proficiency. L2 learners.

Figure 3 shows the proportions of full, lean and omitted Grounds as found in the native adult data. The figure shows a difference between the Chinese data on the one hand and data from the three European languages on the other. The Chinese data show clearly higher proportions of omitted forms (67%) than the European data (around 50% in all three languages). Overall, German is slightly less explicit than the other European languages, given that in addition to omissions it uses lean forms such as darauf, dahinter ('on there', 'behind there') in 15% of the cases (vs. 8% and 7% in English and French respectively).

130

Henriette Hendriks

1 0,8 0,6 0,4 0,2 0

• full

Μi 1 1 English

French

German

• lean • omit

"y

Chinese

Figure 3. Forms used for reference maintenance: Adults Figure 4 allows us to see the development in the LI data. The main finding here concerns the clear increase of explicit forms, particularly between 10 years and adult age. Figure 5 allow us to see the interaction between predicate / location types and explicitness of Grounds in the adult monolingual data. A number of conclusions can be drawn. First of all, comparing the data for cross-linguistics factors, we find the following: changes of location tend to occur without an explicit Ground; in contrast, dynamic predicates expressing a general location tend to attract a high number of explicit Grounds, except in Chinese where only 54% of the situations occur with an explicit Ground. Static predicates expressing a general location attract both full and lean forms, lean forms being more frequent in this type of situation than in any other (less clear for Chinese) (cf. Hickmann, Roland, and Hendriks 1998 for similar results concerning full database). An in-depth look at the development in LI reveals that the development found in figure 4 is due to development in the predicates expressing general locations, rather than changes of location. With the latter type of predicate / location, children from 4 years onward use these more or less as the adults, that is, mainly with omitted Grounds.

Structuring space in discourse English

.

French

!

4 years

131

7 years

ι

ι

10 years

Adults

'

full omit

Γ

4 years

7 years

full

*

~

I

1

10 years

Adults

lean

omit

Chinese

German

1 ~

r~

4 years

7 years

full

*

omit

r 10 years lean

ι Adults

f 4-5 years full omit

Figure 4. Forms used for reference maintenance over age.

1 7 years

1

I

10 years

Adults

132

Henriette

Hendriks

Forms used with dynamic change situations

1 f'T

English

French

German

Chinese

Forms used with dynamic general situations

English

French

German

Chinese

Figure 5. Forms for reference maintenance as a function of situation / predicate types. Native adults

Structuring space in discourse

133

When expressing general locations, however, children of all European languages go from an extreme implicitness at 4 years to a level of explicitness closer to but still not equal to the adult language at 10 years, thereby approximating but not equalling the adult language. The development is very slow, since omitted Grounds remain the most frequent up to 10 years of age. Chinese children form an exception in that they omit Grounds more frequently even with general locations, particularly at 7 and 10 years. Given the overall under-explicitness of the European data, we of course cannot be sure if the pattern of form use with changes of location is indeed adult like, or if it is rather an artifact of a general under-explicitness. When we finally investigate the L2 data (Figure 6), we find that the uses of forms by Chinese learners of European L2's do not vary greatly from one target language to another. The narratives in English L2 show the lowest proportion of omitted forms, narratives in French the highest proportion, German proportions being intermediary. This results in narratives in the French learner variety being clearly more implicit than the target language productions, the Chinese learning German and English adapting well to target language levels of explicitness. Looking at the distribution of forms over the various predicates, we find some more differences. Thus, dynamic predicates expressing a change of location in English L2 are accompanied (in 10% of the cases) by lean forms, whereas no lean forms collocate with these predicates in the native English adults. More detailed analyses reveal that this concerns cases such as (18) in which a deictic verb expressing a change of location into the origo is followed by the deictic expression here, indicating the origo explicitly. Such deictic words do not add much information about the actual location for the listener and seem to be added more out of concern to create a fuller / more complex predicate, rather than to give more spatial information. (18)

The cat come here

134

Henriette Hendriks

Dynamic / change predicates

All p r e d i c a t e t y p e s

ÜJ J chineng

fl

chinfren

full

Ο

lean

Dynamic / general

I chineng

chinger

H

omit

I

full

chinfren ΓΙΙ

lean

chinger §£§

omit

predicates Static g e n e r a l p r e d i c a t e s

chineng I

full

chinfren •

lean

chineng

chinger (3

omit

•

full

chinfren Qj

lean

chinger omit

Figure 6. Forms for reference maintenance as a function of predicate / situation type. L2 learners.

Structuring space in discourse

135

As far as general locations are concerned, Chinese L2 learners of the three European languages are omitting more Grounds when the verbs are dynamic (expressing manner), thereby reflecting their source language habits, as in the examples below in Chinese LI and German L2 respectively. When speaking about static locations, however, they reflect target language patterns in that they provide more full and lean Grounds than found in the Chinese LI data which show a higher proportion of omitted Grounds. As such, L2 learners seem to be reserving verbs expressing manner to talk about manner exclusively, and therefore avoid mentioning location in order not to take the attention away from the manner. (19)

(20)

Täzäi päo he be-at run 'He is running' Die Katz klett (cg0125cat.cod) the cat climb 'The cat climb'

In sum, the types of predicate / location found in the European languages seem very much the same, not only across languages but also across ages, that is, dynamic predicates expressing a change of location are the most frequent in all three languages and from 7 years on. The predicate / location type is an important factor in the explicitness of Grounds as chosen by the speakers. Irrespective of the language spoken, when the predicate expresses a change of location, the Ground is mostly left implicit. This is mainly linked, as we suggested before, to the story itself which takes place around the tree. All protagonist's movements involving changes of location are therefore from or to the tree. Once this location is established as the origo, it can be left implicit. Note, however, that Chinese organise their discourse in such a way that they can rely on the setting when using less explicit forms in the body of the story. That is, they introduce spatial information rather systematically in the setting of the narrative. In contrast, German speakers (children and adults) seem to be overall least concerned with spatial information, and notwithstanding the less "careful" introduction of spatial information, they are also those who maintain reference to space with the least explicit means, English and French speakers being intermediary.

13 6

Henriette Hendriks

5.3. Acquiring the appropriate linguistic means 5.3.1. Complexity of the predicate system in the languages concerned When we discussed language-specific differences as occurring in the four languages concerned here, we hinted to the fact that the four languages analysed in the paper differ significantly in the number of constructions and the degree of complexity of the constructions overall for maintaining reference to space. Note that complexity here concerns the actual number of lexical items contributing to the expression of spatial information, rather than the packaging complexity that we will be discussing below. Earlier, we suggested that this type of complexity may pose a challenge for the learners. We hypothesised that whereas complexity in English and German might provide L2 learners with a less transparent system, the less complex French predicate system might be just as challenging, in that learners might want to express information that is not normally expressed by native speakers. Table 1 in appendix 2 gives an overview of all possible "predicate complexes" that can occur in the target languages, and their appearance in both child LI and adult L2 learner data.6 Concentrating on the adults, in order to assess cross-linguistic variation, we find that the variety of predicate complexes is much more constrained in French than in the other three languages (cf. also Hickmann, Roland, and Hendriks 1998 for a similar discussion of the English and French LI data). As predicted by Talmy and Naigles et al., most predicates in this language consist of bare verb-roots to express changes of location, and verbs accompanied by prepositional phrases to express general locations. In contrast, speakers in the other three languages produce quite a number of clusters of satellites, each cluster slightly different from the next one. Thus, the English linguistic means include predicate complexes such as (21) and (22) in which the same motion event is referred to with a Ground.

(21) (22)

The cat climbs up the tree V + part + Do The cat climbs up to the branch V + part + prep

As we claimed above, Germanic languages, as a result of the number of possible combinations and subtle differences in meaning encoded, are likely to provide a less transparent system for the learner. Some constructions, however, are clearly preferred by the native speaker, such as the choice of a verb plus verb-particle (she flew away) rather than a simple deictic verb (she left) to refer to a change of location (29% vs. 14%). Similarly, when general loca-

Structuring space in discourse

137

tions in dynamic situations are made explicit, they tend to be direct objects (he climbs the tree) rather than prepositional phrases (he climbs into the tree). The German language again provides the speaker with a large variety of constructions (more or less similar to English, except for prepositional adverbs), which is entirely exploited by the adult native speaker. Again, some constructions are clearly preferred in the German adult data, both for the expression of changes of locations, both choices leaving the Ground implicit, as in the other languages (verb + verb-particle, or deictic motion verb + 0). As far as the child data is concerned, we find that, from a young age, the French children tend to construct utterances consisting mainly of verb-roots for expressing spatial information without any elaborations. Additional information (for example about the Ground) is very scarce, and tends to be provided in prepositional phrases rather than in object role. Children learning English and German adapt to the large variety of structures found in the input, so that most constructions are indeed attested from 4 years onwards, reflecting largely similar preferences as compared to the adults. A single difference found in English lies in the choice of prepositional phrases rather than direct object roles for explicit Grounds with general locations in a dynamic situation. Contrary to our expectation that Chinese learners of French might feel the urge to come up with more complex predicates in French than the restricted range offered in the input, we did not find a single predicate that suggests such an attempt in French as an L2 (remember that in their native language they have the possibility of putting verbs in serial constructions, thereby allowing the expression of Manner, vertical direction and deixis (both Path) in one predicate). The Chinese produce very high proportions of "simple" predicate constructions, in which spatial information is restricted to the verb-root (a possibility available in Chinese as well). The L2 learners of English seem no less comfortable with the large variety of constructions in English as the L2 learners of French are with the more restricted range. They use the entire range of constructions, with preferences that are similar to those of the native speakers. In addition, following the tendency of complex predicates in the TL, we find some idiosyncratic constructions in the Chinese learner variety, i.e., some form of serial verb construction, as exemplified in (23) and (24). It is only in English that we find any attempt to produce similar constructions, not found in French, nor German. Finally, the Chinese adults acquiring German as an L2, again make good use of the entire range of possible constructions available in the target language, showing a preference for the same constructions as the adults at all proficiency levels, that is, using verb + verb-particle or deictic motion verbs + 0 mainly for reference maintenance. A difference between Chinese learners of German and German natives is their choice of explicit

13 8

Henriette Hendriks

constructions. Germans will use direct objects, Chinese L2 learners will use prepositional phrases. (23) (24)

and the dog just a run just a runs-follow follow it (ce 15cat.cod) he he he [cat] climb-jump-up (cel7cat.cod)

In sum, both children and adults seem to adapt without problems to the variety of predicate complexes available to express motion and location in the target languages. Children do not construct predicate complexes that are not found in the input. This is more or less what we expected, given that predicate complexity, we feel, is part of the second capacity that children have mastered by the age of four. Adult L2 learners, however, also adjust remarkably well to all three languages. It is interesting to see that they do not produce very complex predicates in French, this target language itself being simple in structure. On the other hand, the English language suggests to Chinese learners that it allows more flexibility, making them add complexity to predicates where it is not found in the LI (adding the deictic expression here to the deictic verb come for example) and attempting (even though infrequently) a Chinese predicate complex, the serial verb construction, as shown in examples (23) and (24) above. We would not want to conclude that English or German target languages are not transparent enough to cope with for the L2 learner, but rather that they provide such a large variety of predicate complexes that the L2 learner feels it is possible to add one more complex.

5.3.2. Complexity in terms of packaging A final analysis concerning the second capacity involved in the acquisition of reference to space in LI and L2 deals with language-specific packaging. For the LI data, we will rely on the findings by Hickmann, Roland, and Hendriks (1998). In this paper, Hickmann et al. compare the development of packaging in English and French, and as such provide the comparison of the acquisition of two typologically different languages, a Romance and a Germanic language (for other languages, cf. Hickmann 2003). Their findings confirm Talmy's and Slobin and Hoiting's description of the typological differences in the native adult data. They furthermore suggest that there is very little development in this domain after 4 years of age. Children seem to be packaging information as in the target language, French children using large quantities of predicates that express Path only or Manner only, English children using predicates which

Structuring space in discourse

139

express mostly a combination of packaged information: manner and deixis, manner and direction, etc. In order to assess the complexity of packaging in the various LI languages, and in their acquisition as L2, Table 2 (see appendix 2) shows the pattern of packaging for the native adult speakers of the respective languages, and the patterns as followed by the Chinese learners of English, French and German. The table shows the following results: In terms of simple vs. complex packaging, the present data show that in French, packaging is simplest (75%), and in German packaging is most complex (only 48% of predicates packages just one element), English being intermediary (55% of packaging is simple). L2 learners seem to follow these tendencies in complexity of packaging. Thus, French adults use 75% of simple packaged predicates, L2 French 90%. Where packaging is most complex (German, 50%), it is in the L2 as well, but note that this tendency is more pronounced at the lower level of proficiency than at the highest level of proficiency. Talmy proposes that, motion excluded, Path is the most basic part of a motion event. If this is indeed the case, we should expect to find Path rather than Manner when packaging is simple. In French this is indeed the case, and simple packaging expresses mainly Path information (DX or DIR) (65%). A similar tendency is found for German (44%). English is the only language which relatively frequently provides simple packaging of Manner (21%). For path not to be expressed at all is extremely rare in German (4%), less rare in French (10%) and most common in English (21%). In sum, it would seem that English concentrates more frequently on Manner only than the other two languages, and is most liberal as concerns the coding of Path in the sense that one finds the highest proportion of motion verbs that do not express path at all. The above analyses concerning predicate complexity and packaging thus seem to show that adult L2 learners handle such complexities very well and adjust to the target language system without much trouble. When one looks at the actual narratives, however, one feels that some important differences between narratives created by adult learners and native speakers do exist nevertheless. It is for this reason that the following more qualitative section is added. One observation that can be made on the basis of Table 2 is that in all learner varieties, the packaging of DX alone (simple packaging) is larger than in the native speaker data. Verb roots in this category are mainly come and go in English, gehen and kommen in German and in French it includes verbs such as venir, aller, partir, sortir, and revenir. These verbs express what motion is most basically about, the coming and going of protagonists. In many cases they are used by the learners instead of verbs that additionally express manner.

140

Henriette Hendriks

If one takes, for example, the event of the dog arriving on the scene (pic. 4), the native speakers tend to frequently use predicates like walk up behind, walk up to, whereas the L2 learners prefer the predicates come, come along, come by. Again, in the situation of the mother bird flying away (pic. 2), native speakers may use fly away, fly o f f , take flight, whereas L2 learners may also use go away, go out or go somewhere else. A rather extreme example of a story following such a simplification strategy is given in (25) below. That this strategy is to some extent related to the lower level of proficiency of the L2 speaker can be nicely shown in example (26) of a German L2 learner of the highest proficiency level, who starts out using a less fully packaged verb, weggehen (Path only), to then replace this verb by the verb wegfliegen (Tath and Manner). (25)

(26)

[...] And eh the baby were very hungry (laughs) IΊI go whether I have some food. Mother said "okay I'm going to catch some foodfor the babies ".And the cat was coming. The cat was very naughty. There's three baby that looks very funny. I want to ask them to come down. He talk to the to to to the swallow: "Shall I come and take you down to the ground? " [...] and he want to climb over the tree and go their nest to come down and he was climbing and climbing when dog come. "Ifyou don 't move down, the baby it will broken [...] (eel lcat.cod) Ja dann geht da fliegen dann die Vogelmutter weg yes then goes there fly then the birdmother away 'Yes, then goes hm ~ there flies hm the motherbird away'.(cgl3cat.cod/proficiency level IV)

Other evidence for a simplification of packaging can be found in the linguistic rendering of the two causative situations occurring with the dog and the cat (dog pulling cat down (pic. 5) and dog chasing cat away (pic. 6)). English native speakers tend to refer to this situation with verbs such as scare away, frighten away, pull down, which express the dog's activity and the causative relation between the dog's and the cat's actions. German adults follow a similar strategy, both languages providing readily available means to do so. French adults choose again to encode all this information in one utterance, even though being forced to use two verbs: faire fuir ('make flee'), since expressions like chasser le chat ('chase the cat') do not express the leaving of the cat. L2 learners of all three languages less frequently package all this information in one utterance. They more often opt for a solution in which the cat is the agent, resulting in renderings of the event as: and the cat ran away; the cat jumped down the tree. When L2 learners do attempt to package all motion

Structuring space in discourse

141

relevant information of that situation, this is what makes them use the resultative verb construction exemplified in examples (23) and (24) above. Finally, whereas we sometimes find this use of simple packaged verbs rather than complex packaged verbs in the L2 learner data, we also find the opposite way of dealing with the Germanic languages. When we look at the climbing situation, then English native adults are very consistent in their choice of predicates: we find climb the tree, climb up the tree and get up the tree. An inventory of the L2 learner list of predicates used in this situation gives us: climb the tree, climb up the tree, climb on the tree, climb over the tree, jump the tree, jump up the tree, across the tree7, and finally climb-jumpup. Similar findings can be reported for Chinese learning German but not for Chinese learning French. This result seems to suggest that the clustering of information in the Germanic languages is slightly less transparent through its wealth, and that, although the Chinese can detect the preferred predicate complex (verb + verb-particle), it is harder for them to detect the preferred verb root and particle used by the native speakers.

6. Discussion The goal of this paper was to get an insight in the structuring of space in discourse by native speakers and child LI and adult L2 learners. Two lines of questions guided this study: 1) questions regarding the universality and language-specificity of reference to space and its structuring in discourse; 2) questions regarding age related aspects influencing the acquisition of reference of space and its structuring in discourse. The combination of reference to space and discourse organisation creates an interesting field of tension between the language-specific and the universal. The spatial situation surrounding speakers may be regarded as more or less universal, leading to the question of the universality of its conceptualisation. The encoding of spatial events cross-linguistically turns out to be highly language-specific, though. Similarly, when structured for discourse, reference to space will have to follow discourse pragmatic principles which, in turn, are considered more or less universal. We thus find ourselves in a domain which is tightly linked to universal principles (spatial and communicative situation) but is known to be expressed in highly language-specific ways. The question resulting from this is: How much cross-linguistic variation are we likely to find given the universality of the underlying conceptual and organisational principles?

142

Henriette Hendriks

The age related aspect ought to allow us to better differentiate between the universal and the language-specific. The particular hypothesis is that in L2 no development should occur regarding universal principles (in contrast to LI acquisition), because learners have already acquired these principles for the first language. Acquisition should be constrained to the language specificmeans for reference to space.

6.1. Cross-linguistic

variation

In the introduction of this paper we proposed that whereas a narrative contains information about person, time, events and space, it relies on the temporal domain for its overall organisation. Spatial information as such does therefore not directly contribute to answering the Quaestio, and can be considered as peripheral. This probably explains why narratives without any explicit spatial information can still be coherent and cohesive. Still, we do expect an adult, in a situation of no mutual knowledge, to provide the listener with a minimum of spatial anchoring. Results show that across languages adults indeed seem to pay attention to the introduction or setting of a spatial frame. They always produce at least one spatial entity that can function as a Ground. Adults also tend to introduce information in the setting of the story rather systematically. And, although less frequently locally marked than reference to person, spatial information is marked for newness in more than 50% of the cases. German adults form the one exception to this pattern. They seem less concerned with the introduction of Grounds in that they are less elaborate in their spatial anchoring, less systematic in their placement of spatial information, and in their marking of newness than adults in French, English and Chinese. Moving on to reference maintenance, we find that once spatial information has been introduced, most following events involve "changes of location" in all four languages. As we observed before, this is mostly a result of the story contents in which frequent changes of location take place. General locations are most frequently referred to by static verbs in Chinese, and by both static and motion verbs in English and German. When general locations are expressed by motion verbs, manner is expressed as well. In French, given that changes of location require a verb packaging path rather than manner, only motion situations with general locations allow these speakers to inform us about manner, which probably explains the higher proportion of this type of predicate in this group overall.

Structuring space in discourse

143

In reference maintenance, we are expecting fewer explicit forms to occur, and indeed in all four languages omitted Grounds form the highest proportions. We find a clear cross-linguistic difference, though, in that Chinese construct narratives with clearly more implicit reference to Grounds than the European languages. This finding corresponds to the more general tendency in Chinese to allow zero forms in reference maintenance and was therefore expected. When comparing within the European languages, German uses slightly fewer full forms and more lean forms than English and French narrators. The kind of predicates used clearly have an influence on the Ground being explicitly mentioned or not, again following more or less universal trends. In all four languages, changes of location appear mostly without explicit reference to the Ground. General locations in combinations with verbs of motion occur overwhelmingly with full forms in the European languages, but only 54% of these verbs are accompanied by full forms in Chinese. With static predicates expressing general locations, French does not allow omissions at all, forms being either lean or full, and this type of predicate attracts most lean forms overall across languages. Chinese, again, is the least explicit language with this predicate. The above results seem to confirm that structuring space on a discourse level is guided by universal principles in that all narratives provide spatial information, provide it preferentially early in the narrative and mark discourse status. However, languages do this more or less consistently, Chinese providing new information in the most elaborate way and earliest in the narrative, relying on discourse (omitting Grounds) in the narrative episode. At the other end of the cline, German adults overall seem less concerned with spatial information. Whereas the difference between Chinese and the Indo-European languages can readily be explained by previously discussed typological differences (topic oriented / subject-oriented, zero licencing, etc.), the differences between German and the other three languages cannot be so easily explained on the basis of available linguistic means. Note, however, that in Hendriks 1993, we found similar patterns for Dutch adults. We then suggested that the lean forms available in that language (forms like dahinter 'behind there' which are less explicit and highly frequent in that language) might allow the speaker to refer to space more frequently in an overall less explicit but no less informative way. Another possible explanation is offered by Carroll and von Stutterheim, who suggest that German spatial information is less structured around objects, and more structured deictically, thereby allowing for less explicit spatial information. When looking at the language-specific means used to realize reference to space, i.e., the means and ways to package and the choice of predicate com-

144

Henriette Hendriks

plexity, the differences in the four languages become very clear. Thus, even though the structuring of space on the discourse organisational level is similar, the linguistic means used are very different. Predicate structure is rather complex in English, and German, as predicted by Slobin, Naigles et al., among others, but not complex at all in French. Similarly, packaging is simple in French, and most complex in German.

6.2. Age related aspects How is this complex system acquired by child LI and adult L2 learners? Starting with the children we find that across languages, talking about space in a narrative is clearly not a priority at 4 years. Although very few stories contain no introductions of Grounds at all, a good number of stories at that age only provide the listener with one possible spatial location. Children become more preoccupied with space at different ages across languages. In English and Chinese children start creating elaborate spatial surroundings at 7 years whereas in French spatial information does not increase until the age of 10. This finding is somewhat surprising in that French adults are most concerned with providing elaborate spatial information. In German and English children follow the adult trend. Children also develop in terms of placement of spatial information in the story (setting/late). Thus, in all four languages spatial information is introduced later in the child data, and earlier (in the setting) in the adult data. It has been shown that newness marking is a late development in children in as far as reference to person is concerned (Hickmann et al. 1996, a.o.). We expect therefore a similar development for the local newness marking of reference to space. The findings confirm these expectations. The appropriateness of newness marking with these referents is lower than in the adult data in all four languages. Moving on to reference maintenance we found that predicates in the adult data mainly express changes of locations. In the Chinese data, this is true at all ages. The speakers of Chinese, from the very early age onwards thus seem to organise their discourse such that spatial information is introduced rather early in the narrative, after which motion is referred to when protagonists move in and out of the set location. In the European languages, children at 4 years use a high proportion of static general locations when narrating, and thus tend to describe what they see more than to temporally relate the actions of the protagonists in order to construct a narrative.

Structuring space in discourse

145

Looking at the forms used for reference maintenance, the children's proportions of omitted Grounds are very high when compared to the adult data (80% overall in the European languages, 75% in Chinese). Looking at the Chinese child data, one might conclude that even 4-year-olds rely on discourse maximally in that they have set a spatial frame and are now deictically moving their protagonists in and out of it. However, the European data contradict this in that settings in those languages are not sufficiently elaborate to allow omitted Grounds, and under-explicitness occurs as a result (attested even at 10 years). Chinese second language learners, rather than reproducing the elaborateness of spatial anchoring as found in their LI, adjust aptly to the level found in the respective target languages, thereby being less elaborate in their L2 than in the LI. When considering the place of introduction in the narrative (early/late), some development is going on in the L2 contrary to our expectations. Where spatial information in the native adult data of all four languages is systematically early (in the setting), this is not the case in French and English L2, but only in German L2. The question is why does this development take place? Several possibilities come to mind, and are probably true for different learners. Not explicitly mentioning spatial locations may be the result of a lexical deficit. It may also be the case, however, that the cognitive load of the narrative task is too heavy for some L2 learners. The obvious way to alleviate this task is to concentrate on the foregrounded information, since this leaves the narrative intact, and to "save energy" on the more peripheral information, i.e., the spatial information. Our guess is that, whereas the second explanation may be valid for all proficiency groups, the first explanation is only valid for very early acquirers of an L2, but this has to be verified in other ways. In reference maintenance, Chinese L2 learners basically use similar predicate / location types as native adults, again adjusting rather closely to the target language use. Concerning the level of explicitness we conclude that, contrary to what was found for reference to person in German L2, no overexplicitation was found in any of the learner languages. In English and German L2, levels closely reflect the target language levels. In French, Chinese L2 learners are even less explicit than the native speakers, thereby resembling results for reference to person in this L2. The fact that French L2 data are even less explicit than the French LI data may be related to the fact that French discourse-organisation as a whole seems to be closer to Chinese discourse organisation than English and German LI discourse organisation (cf. also Hendriks (2002) for a similar discussion for reference to person). French LI seems to have features of a topic-oriented language as well as a subject-

146

Henriette

Hendriks

oriented language (Lambrecht 1981). As far as space is concerned, it does seem to be more concerned with the introduction of spatial information than the other two European languages, whereas more implicit forms occur in reference maintenance, as in Chinese. As a result, Chinese may sense this closeness and tend to construct discourse more like the source language. Finally, we looked at the acquisition of language specific means, and in particular at the adaptation by child LI and adult L2 learners to predicate and packaging complexity. As far as packaging is concerned, earlier studies by Hickmann et al (1996) had already shown that children adapt to the target language system very quickly and mainly before the age of 4, reacting just like adults in their choices of packaging after that age. Adult L2 learners similarly seem to adapt rather well, although a closer look at the data does show some idiosyncratic uses by L2 learners (a slight preference for predicates that package more general (deictic) motion verbs, rather than the more complex packaging of deixis with manner or cause). Some extra attention was given in this analysis to the status of the Path. Talmy (1991) proposes that Path is the more central information with respect to motion, as compared to Manner and Cause. Overall, it could be shown that simple packaging encodes Path more than Manner, and that, if L2 learners chose to stay with simpler packaging, they tend to refer to the Path. The data thus seem to confirm Talmy's claim, except for the English data which allow simple packaging of Manner. Predicate complexity is adjusted to very well also by both children and adults. The French native adult data show a preference for simple predicates, and this is what one finds in the child data as well. The Chinese L2 learners know simple predicates in their source language but also know quite a range of more complex predicates. Faced with the particular target language input, however, they adapt without any problem and use simple predicates mainly (even more frequently than the French natives). Would one be looking at this source-target language pair only, one might say that of course simple predicates are simpler to use. If it were indeed an L2 learner strategy to stick to the provided simpler predicates, we ought to find the same results in English and German L2. But we do not. Where English and German LI predicates are overall more complex, Chinese learners again adapt. In English, they even add to the complexity by using (even though infrequently) idiosyncratic resultative verb constructions. Such complex constructions are not at all found in French L2. Note also, that they occur in a context that is, packaging-wise, complex (causative situations). It is possible that precisely this complexity pushes the learner to be creative and to strive for a more complete linguistic rendering of the situation. At his present level, he choses for a idiosyncratic

Structuring space in discourse

147

means, but similar situations may push him to develop further and acquire the target-language specific means. We started this paper arguing that the construction of an adequate spatial surrounding for events involves three capacities: cognitively representing spatial constellations, finding appropriate linguistic means to express those spatial constellations, and keeping track of the spatial information across utterances in order for the narrative to be coherent with respect to the spatial domain. We made the assumption that the first capacity has been acquired by both child LI and adult L2 learners and should not pose any problems in the experiment, that the second capacity might be problematic for L2 learners but should have been acquired already by 4-year-old children, and that the third capacity, which is linked to universal pragmatic principles, ought to present no problems for adult L2 learners, but should be problematic for child LI learners. There exists a link between language-specific means that influence importantly on the utterance level, and are hard to acquire for the adult learner, and universal principles that operate more on a discourse level, and present a problem for the child LI learner. The fact that language-specific means are used to create universally coherent discourse moreover presents us with an interesting interaction between those two forces guiding language acquisition, the universal and the language-specific. Our data shows this interaction of the two forces already in the native speaker data, in that, even though the four languages under study have very different language-specific means to encode space, they closely follow universal discourse pragmatic principles in their organisation of space. Depending on what aspect one studies, one may therefore have the feeling that the languages work in a very similar way (discourse aspects) or very differently (spatial encoding). As predicted, children take a long time to acquire the structuring of space in discourse, and even the 10-year-olds do not seem to have made this an important part of their narratives. Clearly, constructing discourse is a huge multifaceted task and they can only concentrate on some of the encoded domains. Since spatial information is more peripheral in a narrative, it is the domain that is affected. However, children are very comfortable with the language-specific means available to them for reference to space from 4-years onwards. Against our expectations, adult L2 learners also have some problems with the structuring of space on the discourse level. We feel that the same explanation can be used in this case as for the child LI data, that is, too many domains have to be taken into account at the same time, and the narrators pay less attention to space than to time and person. As a result, they do not show overexplicitation. As far as the language-specific means are concerned, one could

148

Henriette

Hendriks

say that w h e n a target language shows a preference for certain m e a n s / constructions, the L 2 learner will be sensitive to these tendencies, and will in s o m e cases even "exaggerate" them, thereby sounding very close to the target on the one hand, but s o m e w h a t foreign on the other hand.

Notes 1. I would like to thank M. Brala, M. Hickmann and T. Parodi for their helpful comments on an earlier version of this paper. 2. This paper is meant as an overview of work on reference to space by the author within the scope of the Structure of learner variety project. A number of previous papers have already dealt with structuring space in discourse (Hendriks 1993 for Chinese and Dutch L I ; Hendriks 1998a for German and Chinese LI and L2; Hendriks 1998b for French L2; Hendriks and Hickmann 1998 for German and Chinese L1 and L2; Hickmann, Roland, and Hendriks 1998 for French and English LI). In those papers, most of the emphasis was on the introduction of spatial information, and a comparison of the mechanisms with those for the introduction of other types of information (reference to person and objects). In this paper, we will therefore only briefly deal with referent introductions, and refer to the earlier papers when appropriate. Note, however, that data sets in those papers did not involve all three target languages, but in contrast did involve data based on another story (the HORSE story). The choice made here to deal with data of just one story will allow a more overall picture concerning cross-linguistic and age influences, even though it may miss out on some other influencing factors. 3. I'm very grateful to Maya Hickmann for allowing me to use the monolingual database. 4. Cf. Footnote 2. 5. The analyses below exclude all utterances of the narrative that do not convey any information about space, i.e., utterances without any motion, spatial static or posture verbs and without any spatial locations. 6. When the form is attested at all, it is marked by a cross, thereby showing the range of forms that a particular group of speakers use. The most frequent forms (over 10% of the total number of forms) are additionally marked by a circle around the cross, allowing us to see if all groups of speakers have the same preferences for a particular form. 7. Across seems to be used as a verbal element in this example.

References Aske, Jon 1989 Path predicates in English and Spanish: a closer look. Proceedings of the Fifteenth Annual Meeting of the Berkeley Linguistics Society, 1-14.

Structuring space in discourse

149

Bowerman, Melissa and Soonja Choi 2001 Shaping meanings for language: universal and language specific in the acquisition of spatial semantic categories. In: Melissa Bowerman and Stephen Levinson (eds.), Language Acquisition and Conceptual Development, 475511. Cambridge: Cambridge University Press. Carroll, Mary and Christiane von Stutterheim 1993 The representation of spatial configurations in English and German and the grammatical structure of locative and anaphoric expressions. Linguistics 31: 1011-1042. Van Dijk, Teun 1977 Semantic macro-structures aand knowledge frames in discourse comprehension. In: Marcel Just and Patricia Carpenter (eds.), Cognitive Processes in Comprehension, 3-32. Hillsdale, N.J.: Erlbaum. Hendriks, Henriette 1998 Reference to person and space in narrative discourse: a comparison of adult second language and child first language acquisition. Studi Italiani di Linguistica Teorica e Applicata 27: 67-86. Hendriks, Henriette and Maya Hickmann 1998 Reference spatial et cohesion du discours: acquisition de la langue par l'enfant et par l'adulte. In: Merce Pujol Berche, Lucy Nussbaum, and Miquel Llobera (eds.), Adquisicion de lenguas extranjeras: Perspectivas actuales en europa, 151-163. Madrid: Edelsa. Hickmann, Maya, Henriette Hendriks, Fran?oise Roland and James Liang 1996 The marking of new information in children's narratives: a comparison of English, French, german and Mandarin Chinese. Journal of Child Language 23: 591-619. Hickmann, Maya, Franchise Roland and Henriette Hendriks 1998 Reference spatiale dans les recits d'enfants fran?ais: perspective interlangues. Langue Frangaise 118: 104-123. Johnston, Judith and Dan Slobin 1979 The development of locative expressions in English, Italian, Serbo-Croat and Turkish. Journal of Child Language 6: 529-545. Klein, Wolfgang and Christiane von Stutterheim 1987 Quaestio und referentielle Bewegung in Erzählungen. Linguistische Berichte 109: 163-183. Mandler, Jean and Nancy Johnson 1977 Remembrance of things parsed: Story structure and story recall. Cognitive Psychology 9: 111-151. Naigles, Letitia, Ann Eisenberg, Edward Kako, Melissa Highter, and Nancy McGraw 1998 Speaking of motion: verb use in English and Spanish. Language and Cognitive Processes 13: 521-549.

150

Henriette

Hendriks

Slobin, Dan 1973 Cognitive prerequisites for the development of grammar. In: Charles Ferguson and Dan Slobin (eds.), Studies of Child Development, 175-211. New York: Reinhart and Winston. Slobin, Dan and Netty Hoiting 1994 Reference to movement in spoken and signed language: typological considerations. In: Proceedings of the Twentieth Annual Meeting of the Berkeley Linguistic Society, 487-505. Stein, Nancy 1982 The definition of a story. Journal of Pragmatics 6: 487-507. Stein, Nancy and Christine Glenn 1979 An analysis of story comprehension in elementary school children. In: Roy Freedle (Ed.), Advances in Discourse Processes, vol 2: New Directions in Discourse Processing, 53-120. Norwood, N.J.: Ablex. Talmy, Leonard 1975 Semantics and syntax of motion. In: John Kimball (Ed.), Syntax and Semantics 4, 181-238. New York: Academic Press. Talmy, Leonard 1983 How language structures space. In: Herbert Pick and Linda Acredolo (eds.), Spatial Orientation, Research and Application. New York, London: Plenum Press. Talmy, Leonard 1985 Lexicalization patterns: Semantic structure in lexical forms. In: Timothy Shopen, S. Anderson, Talmy Givon, Ed Keenan, and Sandra Thompson (eds.), Language Typology and Syntactic Field Work, vol. 3, 57-149. New York: Cambridge University Press. Talmy, Leonard 1991 Path to realization: a typology of event conflation. In: Proceedings of the 17lh Annual Meeting of the Berkeley Linguistics Society, 480-519.

Structuring space in discourse

151

Appendix I Description of the picture sequences The cat story (1) (2) (3) (4) (5) (6)

There is a nest on the branch of a tree with three baby birds in the nest and a moter bird standing on the nest. A cat has arrived at the bottom of the tree and is looking at the nest. The mother bird is flying away. The cat is sitting down, still looking at the nest. The cat is climbing up the tree. A dog appears in the background. The cat has reached the nest. The dog has grabbed it by the tail. The mother bird is coming back to the nest, holding a worm in its beak. The mother bird is back in the nest. The dog is chasing the cat away.

152

Henriette Hendriks

Appendix II

(Ν J

3 -α
1

α

S 3 & xi ω >

ο ο U Sü ω i: Ό u + (-£ J3

-α

+

J2 u ω

a3 Ο c ο Ο* "5 3 Ο C

C ο ω

£

+

.ο

t:

£

.a ^ £ f-

Structuring space in discourse

+

c3 Ο cο ίο* "a 3 Ο C + c ο ο QΟ —

+

c ο

Ο α, ο + -Ω

Ο. +

,S ffο — & U (U >

+

1 β-Η Ο. Ci. -Ω ι- a >

+ >

3 Q

"β

1

C 3 ρ

Λ ο

6ο -a .c £ ε -ο 3 ε + -Ο

00

•S

s

c .ο

+

C ΧL Ο Ο ω 'ίΛ ο α Ο •JE ο Ο. 1> Έ a -α + + •Π Ιt- "< Κ3 -S Χ) ο > ΰ. >

£

153

154

Henriette

Hendriks

>

top foe Aux Mod Vlex

Other Symbols enclose a topic or focus string topic focus temporal auxiliary modal auxiliary (vouloir, pouvoir, falloir) lexical verb Some definitions

Anaphoric negation (in different cases from those of a prophrase)

Categorial negation Holoprastic negation or prophrase

Relational, functional or non-thematic verbs Thematic verbs

negative reaction in which non is followed by some information drawn by the preceding context negation that bears on the category referred to by a referential NP negative reaction limited to the use of non, whose function is that of an anaphoric element

copula, avoir and temporal or modal auxiliaries lexical verbs

400

Patrizia Gialiano and Daniel

NS: MB: MH: ZA: AB: BE: PA: AL: Glo:

Vironique

Native speaker interviewer Malika B. Malika H. Zahra Abdelmalek Bernarda Palmira Alfonso Gloria

References Adila, Aziz 1996 La negation en arabe marocain (le parier de Casablanca). In: Salem Chaker and Dominique Caubet (eds.) La Negation en herbere et en arabe maghrebin. Bulletin des etudes africaines 1: 99-116. Alarcos Llorac, Emilio 1994 Gramätica de la lengua espanola. Madrid: Real Academia Espanola. Arrive, Michel, Fran