Appel à projets franco-allemand en sciences humaines et sociales
Ausschreibung eines deutsch-französischen Programms in den Geistes- und Sozialwissenschaften
ohne thematische Vorgaben
Titre du projet
Structure de l’énoncé en contexte :
Acquisition des langues premières et secondes dans une perspective inter-langues
Utterance structure in context :
first and second language acquisition in a cross-linguistic perspective
Äußerungsstruktur im Kontext. Erst- und Zweitspracherwerb in sprachvergleichender Perspektive
Utterance structure in context :
first and second language acquisition in a cross-linguistic perspective
Acronyme : LANGACROSS
Kennwort : LANGACROSS
Noms des deux coordinateurs
(français et allemand) du projet commun et de leurs institutions
Laboratoire « Structures formelles du langage » (UMR 7023 CNRS & Paris 8)
Max-Planck Institut für Psycholinguistik
(auf deutscher und auf französischer Seite) und ihrer Institution
Max-Planck Institut für Psycholinguistik
Laboratoire « Structures formelles du langage » (UMR 7023 CNRS & Paris 8)
Présentation du projet
1. Fiche d’identité du projet
(also see Table 1)
Laboratoire « Structures Formelles du Langage » (UMR 7023, CNRS & Paris 8)
2, rue de
Max-Planck Institut für Psycholinguistik, Nijmegen
· Members: D. Bassano, S. Colonna, B. Copley, M. Lambert, K. Ferret, C. Perdue, P. Taranne, M. Watorek
· Doctoral students: T. Aleksandrova, A.-Cl. Demagny, N. Ghantous, A. Kapral, P. Leclercq, A.-K. Ochsenbauer, I. Saddour, E. Soroli
Ø French Team 2: Laboratoire « Savoirs, textes, langages » (UMR 8163, CNRS & Lille 3), BP 60149, F-59653 Villeuve d’Ascq, France
· Members: E. Mathiot, S. Benazzo, M. Lemmens
· Doctoral student: Mark Tutton
Ø German Team 1: Max-Planck Institut für Psycholinguistik (PB 310, 6500 AH Nijmegen, The Netherlands)
· Members: A. Chen, M. Gullberg, W. Klein, B. Narasimhan, L. Roberts
· Doctoral students: L. Herbst, S. Schimke, G. Stanic, J. Verhagen, M. Ellert
· Members: M. Carroll, E. Kaltenbacher, Ch. von Stutterheim, N. Tomita
· Doctoral students: D. Boit, A. Bouhaous, G. Carpena, N. Castro Alonso, M. Essam, M. Flecken, J. Chuang, J. Ge, S. Natale, N. Sahonenko
1.2. Title, acronym, key words
1.3. Disciplines and scientific domains
Language acquisition, language structure, language typology
1.4. Duration of project
Language acquisition is an intrinsic aspect of human behaviour, both a necessary stepping stone in ontogenetic development and a vital task in our multilingual world. Despite an increasing amount of research, fundamental questions concerning the mechanisms of the acquisition process are still sharply debated. Different frameworks strongly diverge in many respects. To name but three: is the language faculty domain-specific and innate or the product of a gradual development through general learning processes? What is the impact of typological constraints on the acquisition process? How important is the role of language use in context for learners’ developing linguistic competence?
Our aim is to address some of
these questions in three language domains that have been at the centre of most
debates, examining in each case both the sentence and discourse levels of
linguistic organization: 1) information structure, 2) time, and 3) space.
Particular attention is paid to the role of universal cognitive vs. language-specific,
and structural vs. functional determinants of acquisition. This aim requires a
two-fold comparative perspective – of learner type; of linguistic sub-systems.
We compare the developmental progressions followed by two types of learners:
children - whose cognitive system is developing during first language
acquisition - and adults - who are cognitively developed when acquiring a
second language. Cross-linguistic comparisons examine the acquisition of a
number of linguistic systems that display relevant contrasts in relation to
particular claims. Participants belong to four European teams, two in France
(UMR 7023 & 8163) and two in
2. Scientific context
2.1. The field of language acquisition
Language acquisition is a major domain of study among the cognitive sciences. Research in this domain stems from different disciplines and scientific traditions, including philosophy, linguistics, psychology, and the neurosciences. Depending on the discipline, on the theoretical framework adopted, and on the particular issues at stake, this research has focused on different phenomena (morpho-syntax, semantics, pragmatics, phonology) displayed by different types of learners and acquisition situations (first vs. second language acquisition, simultaneous bilingualism) and conclusions diverge with respect to several fundamental issues. For example, major claims have been made with respect to the role of several types of determinants during language acquisition: 1) universal cognitive vs. language-specific determinants, and 2) structural vs. functional determinants. As shown below, these questions are far from having been resolved so far. The aim of the project will be to assess these claims in the light of available data bases concerning both first and second language acquisition across languages, and to add to these data bases where necessary.
What the project does not set out to do, however, is engage in
the so-called ‘logical’ problem of language acquisition. In the tradition of
generative grammar, it is assumed that the human language faculty essentially
consists of a set of universal organising principles (“Universal Grammar”: UG)
which define the
· the role of cognitive and communicative capacities of different types of learner in the acquisition process;
· the way these capacities interact with the structural properties of the language to be learned;
· the successive stages in the acquisition process;
· how and why a learner passes from one stage to another.
In relation to previous work, the project expects to further our understanding of the acquisition process in three overarching areas, and has furthermore the practical aim of broadening and systematising available data bases. As shown in more detail below, the first contribution of the proposed project is to compare the processes whereby children acquire their native language and adults acquire a second language. One can view the language acquisition process as a complex interaction between three families of determining factors :
(1) the cognitive and perceptual capabilities of the learner ;
(2) the linguistic environment in which the learner evolves ;
(3) the learner’s communicative needs and motivations.
These factors are given a remarkably different weight by competing theories of acquisition, and intervene in very different ways depending on the type of learner - child vs adult as a first approximation. One of the striking differences between child and adult acquisition – and a difference that warrants explanation – is the result of the process. In marked difference to the child, the adult rarely achieves full native mastery of the L2, and the end result of the process shows great individual variation. Here again, different theories give different weights to factors (1) to (3) above in attempting to explain this « fossilisation ». The adult is considered to be a learner who is “cognitively developed when acquiring a second language”. His/her learning task is therefore usually considered, including in our own previous work, to be to find new linguistic means to express (fully developed) available concepts.
What has received much less attention is that in fundamental conceptual domains such as space and time, the grammar of a language expresses conceptual distinctions presently unavailable, or only partially available to the adult learner. Such linguistic distinctions may therefore lead to a (partial) reconceptualisation of the domain, with the consequence that the adult’s learning task is not exclusively linguistic. The failure to take into consideration such new cognitive distinctions may explain some aspects of adult “fossilisation”, and leads to a necessary reassessment of the “universal/ specific” distinction. Similarly, in first language acquisition, cross-linguistic research has begun to raise fundamental questions concerning the impact of language-specific properties on how children select, organize and reorganize information during the course of development, leading to conceptual systems that partly differ across language groups.
These questions will be examined in several language domains, which are at the centre of current debates concerning the nature of language and the process of language acquisition: 1) information structure, 2) time and 3) space as they are expressed by utterances in context. In addition, two “emergent themes” which necessarily involve looking at aspects of all three domains will be addressed. These are (a) the problems of building up a conceptual representation (or, for the adult, modifying an already available conceptual representation) as described in the previous paragraph, and (b) the interrelations between the three domains of investigation during the acquisition process. These themes will be described in paragraphs C1 and C2 of 3.1 below.
Domain I: Information structure
With respect to the earliest stages of first language acquisition, much literature concerning information structure has been devoted to how children mark sentence-internal relations from a grammatical, semantic and/or pragmatic point of view (subject/object, agent/patient, topics). Classical “cognitivist” models in psycholinguistics (e.g., Brown, 1973; Clark, 1973; Slobin (ed.), 1985) postulate that all children start out with cognitively salient “canonical” schemata that guide language acquisition. In such views, children first construct cognitive representations during the pre-linguistic stage and then search for the means available in their surrounding linguistic system to express them, relying on universal perceptual strategies.
For example, research shows recurrently that particular types of semantic information are expressed by children in a considerable number of languages, suggesting the existence of salient scenes such as the disappearance of concrete entities (“all gone”) or Agent-Action-Patient scenes (NVN utterances) involving an entity (agent) that acts (action) on another entity (patient) with a resulting state (the patient undergoing the event is affected in some way). Furthermore, according to one of Slobin’s “operating principles” (“Pay attention to the end of words!”), the acquisition of morphological markings should be guided by a universal perceptual strategy that leads children to concentrate their attention on information that is situated at the right boundary of open-class elements where much relevant personal, temporal, spatial information can be found in verbal and nominal morphology (finiteness, subject-verb agreement, person, number, gender, case).
In contrast to children, adult learners are not sensitive to morphology. There is general agreement that learners first acquire open-class items which are juxtaposed in simple utterances, then closed-class items are added to the repertoire, and morphology is a late development, if acquired at all. As Parodi, Schwartz and Clahsen (1997:2) put it: a “descriptive generalization on which L2 acquisition researchers seem to have achieved agreement is that inflectional morphology poses major acquisition problems for adult L2 acquirers”. This implies that the target languages’ way of signalling basic information structure distinctions, e.g. in the domain of temporal relations develops only gradually in adult L2 learners. Bartning (1997) develops the acquisitional sequence in great detail, proposing detailed linguistic descriptions of five successive stages. The development of finiteness marking (Schlyter, 2003; and the interrelation of this development with other phenomena: negation (Meisel, 1997; Parodi, 2000), word order, person and case agreement (Herschensohn, 2001) has proved to be a recurrent domain of research.
there is controversy is in explaining the sequence. The characterisation of the
Domain II: Time
In the temporal domain, a considerable amount of research on first language acquisition has focused on the hypothesis that children first acquire verbal morphology on the basis of conceptual representations that privilege results (see Hickmann, 2003 for an extensive review). According to this “defective tense hypothesis”, children first use tense morphology to mark the immediate results of events that have occurred in the here-and-now, rather than to mark temporal relations per se, thereby first associating past markings to predicates that represent changes of states. The prediction then is that children’s first past marking should be of a particular type (perfective past) and that they should be used in relation to predicates that present particular semantic properties (bounded or telic).
Although this hypothesis has received considerable support in developmental data from a large number of languages, it has also been debated on the basis of three dimensions. First, some cross-linguistic developmental data does not fit the predicted patterns, for example data from languages that present a particularly transparent verbal morphology (e.g., Polish, Weist et al., 1991). More generally, the acquisition of morphology has been shown to depend on a number of typological dimensions, such as the extent to which morphological systems are rich, transparent and uniform (Bittner et al., 2003). A second debate concerns the age at which the predicted pattern should occur. In particular, although the original claim founding the defective tense hypothesis (Bronckart & Sinclair, 1973) was based on experimental data elicited from children of 4 to 10 years, the claim has been extended to the earliest stages of morphological acquisition. Finally, a major debate concerns the extent to which the discourse functions of morphological tense-aspect devices can be taken into account by the defective tense hypothesis. It is now well established that temporal-aspectual devices play a universal role in discourse organization as universal markings of information grounding. All other things being equal, perfective devices mark the discourse foreground, while imperfective ones mark the background. In this respect, the defective tense hypothesis merely focuses on semantic content at the utterance level, paying little attention to pragmatic aspects of children’s utterances. Recent studies carried out by the project partners indeed raise fundamental questions in this respect, to which we return below.
Given the acquisitional sequence sketched above, the first explicit markers of temporal relations to emerge in adult acquisition are adverbs. Most international work in this area has nonetheless been exclusively devoted to relatively late markers of temporal relations, namely verbal morphology. As in L1 acquisition, the defective tense hypothesis (Andersen & Shirai 1994) has given rise to many studies (e.g. Housen, 2002). Other researchers (Huebner 1983, Meisel 1987, Sato 1990) specifically investigated past time reference. Sato observed a complete absence of regular past verbal inflexion in her 15 month study of two Vietnamese boys, Meisel found that his subject, José, initially used verbs in an invariant form, and the acquisition of tense involved many overgeneralisations. He also notes that many informants never used anything which came even anywhere close to the German system of inflexion. What needs to be explained therefore is why this should be so (see the ‘end-state’ below).
Similarly, in the spatial domain, the most predominant view is that learners’ linguistic expression of space involves basic universal cognitive and perceptual capacities that serve as prerequisites for their acquisition of spatial language. Thus, the acquisition of spatial prepositions (see an extensive review in Johnston, 1988a, 1988b) has been claimed to follow universal stages based on cognitive determinants, such that – regardless of language- learners first acquire neighbouring and containment relations (near, in), then relations along the vertical axis (first on/under, then above/below), then relations along the sagittal axis (in front of, behind). More generally, it has been suggested (Landau & Jackendoff, 1993) that languages make a universal distinction between two sub-systems, one dedicated to referring to entities and the other to localising them (what/where systems), reflecting two different neural systems in the brain.
However, more recent cross-linguistic studies (Bowerman, 1996, Bowerman & Choi, 2003; Choi & Bowerman, 1991) suggest a more complex developmental picture that casts such universal principles into doubt. Results suggest that children’s productions are more similar to those of adults speaking the same language than to those of children of the same age speaking a different language. In similar vein, adults adapt relatively early to the L2 system and show little transfer in this area (Becker & Carroll 1997). Furthermore, much research (Slobin, 1996a, 1996b, 2003, 2004, 2006; Talmy, 2000) has shown considerable variation across spatial systems that go well beyond the expression of spatial relations. Thus, the expression of motion events shows a major typological distinction between two language families, depending on the ways in which spatial information is lexicalised or grammaticalised. In satellite-framed languages (Germanic) verb roots typically express manner and verbal satellites, whereas in verb-framed languages (Romance) they typically express path (e.g., He ran across the garden vs. Il a traversé le jardin).
These language families differ in how they express varied information beyond the spatial domain, for example in the expression of causal relations. Thus, compact causative-resultative constructions in Germanic languages expressing together cause, manner and path (e.g., She blew her hair dry, She kicked the door open, The wind blew the leaves away) are not available in Romance languages (Elle s’est séché les cheveux, Elle a ouvert la porte à coups de pied, Le vent a soufflé sur les feuilles qui se sont envolées). These different paradigms strongly influence speakers’ productions, including at the youngest age, from the emergence of language on. Such results have lead to more “relativistic” models, according to which languages provide filters that channel information in particular ways, leading children to pay more or less attention to some aspects of reality, that thereby become more salient and accessible.
2.2. Previous research of partners
The project brings together teams with an established track record of joint work, and who have already made significant contributions in the field of language acquisition. The researchers in these teams have worked on children’s first language acquisition or on adults’ second language acquisition, but they have addressed similar questions concerning the central theoretical framework of the project.
A. First language acquisition
Much of the participants’ work has been devoted to the major constituents of utterance structure, namely nouns and verbs: (i) the development of NPs, (ii) the interrelation of utterance structure and the discourse context, (iii) verb-argument structure, (iv) the emergence of verbal forms, and (v) spatial relations in discourse.
In a series of studies Bassano and collaborators (Bassano, 2000; Bassano & Eme, 2001; Bassano, 2005; Bassano, Maillochon & Mottet, 2007) have investigated when and how French-learning children acquire the main grammatical constraint on the noun category, i.e., the obligatory use of a preceding determiner. Results indicate that noun grammaticalization in French children is a gradual process which involves the use of transitional structures, such as early pre-nominal fillers, as well as an increasing diversity in the content and context of determiner use. This process is influenced by both prosodic factors, such as noun length effects, and lexical factors, such as animacy. Prosodic factors predominate in the first steps of the developmental process, when monosyllabic nouns are more likely to take a filler or determiner. Lexical influences predominate in later stages (animacy).
(ii) Utterance structure and discourse context
Hickmann and collaborators (Hickmann, 2003 ; Hickmann & Hendriks, 1999; Hickmann et al., 1996) have studied discourse organization on the basis of narratives collected across several languages (English, French, German, Chinese). Results highlight the joint impact of universal (cognitive, pragmatic) and language-specific factors. With respect to information structure, results show that in all languages examined 1) children first mark new information at around 6-7 years, 2) they first do so with local means (nominal determiners), and 3) from 4-5 years on they follow discourse pragmatic principles governing the local marking of given information (pronominal forms as a function of coreferential relations). Differences across languages, however, also show wide variations of clause structure as a function of grammatical constraints (e.g., topic-drop in German vs. obligatory subjects in French, obligatory V2-rule in German, obligatory preverbal clitics in French).
Research using the intermodal preferential looking paradigm (Kail, Boibieux, Coulaud, 2005) has investigated very young children’s comprehension of transitive and intransitive constructions. Results show that these constructions are understood at around age 24-27 months, with a slight advantage for transitive constructions. However, it is not clear what representations underlie these results, e.g., whether children integrate information of a semantic nature, in line with the “verb-island” hypothesis (Tomasello, 2000).
(iv) Temporal markings: forms and functions
With respect to temporal markings, Bassano, Maillochon, Klampfer & Dressler (2001), and Bassano, Laaha, Maillochon & Dressler (2004 ) closely examined the emergence of verbal forms in a cross-linguistic study of the spontaneous productions of German- and French-speaking children. The acquisition of periphrastic constructions clearly underline the interaction between linguistic and cognitive constraints on the young child: it was shown that the periphrastic future was late in emerging in both languages, whereas the superficially similar passé compose/ Perfekt form showed differences. Acquired early in French, it emerged much later in German. One may hypothesize that the different aspecto-temporal functions of these forms in each language affected acquisition rate. A further question to be investigated was the relationship between lexicon and grammar in the development of verbal morphology. In four case studies, developmental analyses showed precursors, such as bare infinitives and past participles and pre-verbal fillers, which indicate a gradual acquisition process. They also showed relations over time between grammaticization and lexical production of verbs, which is an argument for the ‘critical lexical mass’ hypothesis and for interdependencies between lexical and grammatical developments.
Some of previous results on discourse organization also tested the defective tense hypothesis (Hendriks, Hickmann & Liang, 1999; Hickmann, 1996, 2003). The results only partially confirm this hypothesis. Thus, they do show the predicted relation between predicate types and perfective markings in all languages (bounded verbs and past perfective verbal inflections or aspect particles). However, the strength of this relation varies considerably as a function of age and language. Thus, this relation is strongest in Chinese at all ages and it is also strong in English, but only at the youngest ages, whereas it is quite weak in German and French (at all ages). These differences are linked to the different types of temporal anchoring adopted by speakers across languages. In this respect, regardless of these language differences, tense-aspect alternations take on discourse functions from 6-7 years on, serving to mark grounding relations in cases of simultaneous events. In addition to verbal markings, children gradually rely on temporal-aspectual adverbials to mark discourse functions, although they do so earlier and more heavily in Chinese (where there is no tense morphology) than in the other languages.
(v) Locations and changes of locations in discourse
Finally, strikingly similar developmental progressions show children’s gradual ability to provide spatial anchoring (locations) at the beginning of their narratives, allowing the interpretation of subsequent discourse (changes of location). In all languages, this ability emerges at about 6-7 years and continues to develop after 10 years until adulthood. However, the expression of motion events shows considerable variations across languages that follow from the typological properties discussed above. Thus, children use denser and more varied motion predicates in German (satellite-framed language) than in French (verb-framed language). In particular, German children express manner, path and/or cause in compact constructions, whereas French children focus mainly on path in cases of changes of location. Subsequent experimental results confirm these results in comparisons of French with English (Hickmann, 2006, 2007; Hickmann, Hendriks & Champaud, 2007) or French with German (Ochsenbauer, 2006; Ochsenbauer & Hickmann, in preparation).
B. Second language acquisition
Joint work on second language acquisition between many members of the participating teams started in the 1980s with the European Science Foundation’s project « Second language acquisition by adult immigrants », coordinated by Klein and Perdue, and whose results are summarised in Perdue (1993). Three major research themes of the project were Utterance structure (Klein & Perdue 1992), the expression of temporal relations (Dietrich, Klein & Noyau 1995) and the expression of spatial relations (Becker & Carroll 1997). The methodology was longitudinal and cross-linguistic. Ten different pairings of native-target languages were studied, the targets being Dutch, English, German, French, and Swedish. Data are available for 40 longitudinal case studies over 30 months. The approach adopted in this and subsequent work sees the process of second language acquisition not in terms of errors or deviations from a target, but rather in terms of the two-fold systematicity it exhibits:
· the learner variety is a system in its own right, error-free by definition, and characterised at a given time by a particular lexical repertoire and by a particular interaction of organising principles ;
· the dynamics of the acquisition process can be found in the interaction between information structure and the linguistic means the learner has available at any given time, i.e., the way the repertoire is structured so as to allow (or not to allow) the expression of a complex body of information in context.
An underlying assumption is therefore that development in the direction of the morpho-syntactic specifics of the target language takes place when the available linguistic means do not allow the learner to cope with specific discourse contexts. This assumption puts the work firmly in the perspective of the developmental – as opposed to the ‘logical’ – approach to language acquisition (see above 2.1).
The participants’ relevant prior work has focused on the four following areas of interest.
(i) Finiteness and Information Status (topic-comment-organisation)
The structural and functional role of finiteness phenomena is reflected in language acquisition. Second language learners typically start with nonfinite forms and only then slowly acquire finiteness, a process which leads to a substantial reorganization of their sentence structure. It was found found that constituents are first related by adjunction, while their ordering is based on principles of information structuring. Furthermore, the relation between the topic and the predicate, i.e., the nonfinite component of the utterance, is established by a closed class of lexical phrases expressing illocutionary force. These "Illocutionary Phrases" allow learners to express "volition", "ability", "possibility", "obligation" and "contrastive.assertion", while the expression of ordinary "assertion" occurs by default (Dimroth et al., 2003).
Similar principles of information structure related to the semantic and pragmatic knowledge base of the learners were shown to be responsible for the placement constraints of negation operative at different levels of acquisition (Becker, 2005).
(ii) Scope (focus particles, negation)
Prior work on the acquisition of scope bearing elements (scope particles, temporal adverbials, negation) in elementary learner languages (e.g. Andorno, 2005; Becker, 2005; Benazzo, 2003; Dimroth, 2002; Watorek & Dimroth, 2005) has demonstrated how first and second language learners learn to integrate such particles in utterances of growing structural complexity. These studies have focused on the question how adult learners of different target language systems mark the scope of these different types of particles when progressing from an elementary non-finite to a more developed finite utterance structure. Correponding investigations into L1 development in different languages have shown that children have less difficulties with the structural integration (position relative to scope) of such items (Nederstigt, 2003). They employ at least a sub-set of the target languages’ integration possibilities in an adult-like way, but show considerable deviations on the conceptual level, where they only gradually tune into the language specific ways of expressing additive relations to ensure discourse coherence (Watorek (ed.), 2004).
(iii) Temporality and the construction of complex discourse
Language specific effects in the selection and structuring of information in advanced learner languages were studied in crosslinguistic analyses, addressing the difficulties that arise in reaching nativelike proficiency when structuring the semantic domains of time, events, space, entities, etc. in L2 discourse. Many factors that distinguish very advanced learners from native speakers could be ascribed to the way in which grammaticized categories relate to patterns of information organization in the learners’ L1 linguistic systems. A comparison of film retellings by native speakers of English and German revealed significant crosslinguistic differences in the events selected for mention and the perspective taken in temporal frames of reference: German speakers present events holistically (i.e., events are represented as bounded and thus include an endpoint), while English speakers tend to decompose events into different phases, preferring a temporal perspective that incorporates ongoing, unbounded events. (Carroll & Lambert 2003; von Stutterheim & Lambert 2005).
C. Systematic comparisons of first and second language acquisition
Some first attempts at comparison between first and second language acquisition have been undertaken across these teams (Dimroth et al., 2003; Gretsch & Perdue, 2007; Hendriks, 2005, Hendriks & Hickmann, 1998, Watorek (ed.) 2004) and such comparisons are to be expanded in the present project, where the comparative dimension of acquisition at different ages will be systematically investigated. What is intended in the present project is (a) to compare this work more systematically with results from first language acquisition, and (b) further to investigate the interrelation of these phenomena. These two points are taken up again in the discussion of the project’s objectives below.
3. Aims and program
The project would provide a unique opportunity for the participating teams, building on previous collaborative work, to obtain a coherent, synthetic picture of the cognitive-linguistic interplay in the acquisition processes of different types of learner, put to the test of a wide range of languages. The teams have already created rich data bases, which will be added to selectively in order to test specific hypotheses across languages. Of particular interest from this point of view are the “emergent themes” of the interrelation of spatial, temporal and personal reference with informational constraints in context, in the building up of a linguistic repertoire (see C1 below), and the building up of a conceptual representation of cognitive domains such as space and time during the acquisition of the specific linguistic means for their expression which a particular language offers (see C2 below).
A. scientific advances expected from this French-German cooperation
Given the previous contributions of the participants, cooperation among the French and German teams will be particularly fruitful in promoting collaboration in the following language domains, which have been investigated to a greater or lesser extent by all the participating teams: 1) information structure, 2) time, and 3) space. Within each of these overarching domains, the particular topics to be addressed are shown in Table 2. The project will structure the research that has been partially undertaken under these different entries by providing a strong and coherent inter-disciplinary framework that brings together researchers from different scientific traditions (psychology, linguistics), all concerned with a comparative approach to language acquisition. As shown below, our contribution stems from a comparative approach that combines two dimensions: different languages and different types of learners and developmental processes.
In addition, it is planned to investigate (a) the interrelations between these three domains during the acquisition process (the expressive means built up for the temporal or spatial domain have an impact on the way this information is distributed in the utterance, and so on), and (b) the way conceptual representations are built up by the learner, in interaction with the grammar of the language being acquired. Much previous work by all teams has indicated that these “emergent themes” have great potential for future research: what is now needed is a specific and synthetic approach which has hitherto not been possible. These themes are discussed below in C1 and C2.
Research questions to be addressed in the project
Utterance structure in context:
first and second language acquisition
in a cross-linguistic perspective
1. Givenness (accessibility)
2. Topic – assertion – comment
3. Contrast and additive relations in discourse
1. Time and topic-focus structure
2. Temporal relations in discourse
3. The grammaticalization of time
1. Space and topic-focus structure
2. Static spatial relations
3. Lexicalization and grammaticalization
of motion events
With these studies, we expect to contribute significantly to the study of language typology from a psycholinguistic perspective, and to increase our detailed understanding of different types of developmental process, this latter aim having potential importance for language pedagogy and remediation. The « target » languages to be investigated are mainly, although not exclusively, Romance and Germanic (with special attention to French and German), although other comparisons will be possible in some areas (Slavic, Semitic, Asian languages; see data bases below). This obviously represents a practical choice for a Franco-German project, but much more interestingly, these languages show systematic cross-linguistic variation in the relevant areas of investigation.
Domain I: Information Structure
The project aims to study the linguistic realization of aspects of information structure and its acquisition by children and adults learning different languages. While the semantic and syntactic properties of linguistic phenomena such as word order and intonation have received considerable attention in native languages, less attention has been paid to their interaction with information structure and how it is manifested in learner language. The following three dimensions of information structure and their expression in learner language will be addressed in turn:
1. Givenness (accessibility)
2. Topic – assertion – comment
3. Contrast and additive relations in discourse
Speakers tailor their utterances to meet the informational needs of the hearer in a number of ways. For instance, an entity that has been mentioned before in the discourse is more “accessible” in the minds of speakers and hearers than an entity that has never been explicitly mentioned. The speaker might distinguish between accessible and inaccessible referents using (one of) several linguistic means, including word order, intonation or nominal forms such as unstressed pronouns (it) or definite NPs (the boy). Participants in this part of the project plan to address the following research questions:
1. What is the role of information status in childrens’ word order preferences? (Dimroth, Narasimhan, Hickmann)
These research questions will profit greatly from the cross-linguistic comparison envisaged in this French-German co-operation. In particular, further research on learners’ acquisition of NPs and the morpho-syntactic constraints on determiner uses will expand previous results to cross-linguistic comparisons between French and Germanic languages, which present different prosodic, lexical and syntactic constraints of determiner uses. Further cross-linguistic comparison is possible with Polish, which, in the absence of a determiner system, signals givenness by word order and other lexical means. In addition, further research on children’s acquisition of transitive constructions will aim at capturing the developmental progression allowing children to pass from concretely oriented representations to more abstract structures. Special attention will be placed on the respective role of French nouns and pronouns in the comprehension of these structures while using the French-specific possibility of associating these categories in dislocations whose function is to thematise the referent. Results will be compared to the comprehension of other languages (e.g., null-subject or topic-drop languages) and to findings from adult bilinguals, focusing on the question, what the role is of such L1 specific associations for preferences in sentence processing (Roberts).
Finally, research will examine young children’s productions in Romance and Germanic languages to determine how the discourse functions of different NPs (nouns, pronouns, ellipsis) emerge in discourse context, in relation to clause structures (dislocations). This set of research studies will include monolingual as well as bilingual children, and will greatly benefit from collaboration integrating results concerning adult L2 learners carried out on similar topics.
Topic - assertion – comment
When speakers attempt to update the hearer’s mental model by providing information about another entity, property or event, they must indicate the partitioning of a sentence into a topic (what the sentence is about), a comment (the information provided about the topic) and the type of semantic relation between the two (assertion, negation etc.). This can be marked linguistically in a variety of ways, including intonation, word order, and discourse particles. Team members aim at investigating the following questions:
1. Assertion and Finiteness L1 and L2 acquisition (Jordens, Perdue, Ferret, Verhagen Benazzo, Dimroth)
To date, most of the participants’ joint work on the acquisition of information structure and assertion marking has focused on the L1 and L2 acquisition of Germanic languages (Dutch and German, in particular; Dimroth et al. 2003; Jordens & Dimroth 2006, Gretsch & Perdue, 2007, Verhagen, 2005). With the PhD thesis of Schimke (in prep.) the first controlled comparison between German and French is under way. Based on current proposals by Jordens, 2007 and Ferret & Perdue 2007, it is planned to further develop this cross-linguistic comparison with French.
Contrast & additive relations in discourse
Building on prior work by project participants, this part of the project focuses on investigating preferred patterns of distribution and marking of contrastive vs. maintained information. Recently, the focus of interest has shifted towards the interaction of such particles with the information structure of utterances in connected discourse. Participants of the relevant teams plan to study how L1 and L2 learners of different languages mark contrastive and additive information in the domains of times and entities and how they express polarity: contrastive assertion and negation). The following research questions are at the centre of interest:
1. Additive/temporal scope particles and the L2 acquisition of finiteness in a developmental perspective (Benazzo, Schimke, Verhagen, Dimroth)
2. L1 and L2 learners’ acquisition of language specific preferences for marking contrast and addition across time and entities (Andorno, Becker, Benazzo, Dimroth)
3. The intonational marking of information structure and scope relations in child language (Chen, Stanic)
Cross-linguistic data collected with the help of a joint elicitation tool (video-clip “finite story”, MPI, 2005) reveal that native speakers of different target languages differ in their preferences for verbalizing situations in which similar events apply to different topic entities (protagonists in narrative discourse) at different time intervals. While speakers of some languages (e.g. French) predominantly chose the explicit marking of the addition of temporal intervals (continuation, repetition, restitution), others (e.g. German and Italian) tend to leave these temporal additive relations unmarked and prefer to express, that similar events hold for new topic entites in addition to others. Considerable language specific differences are also found in the available repertoire of particles marking contrastive assertions / polarity (e.g. German doch, Dutch wel) and the combinatorics of such particles (e.g. German doch auch noch nicht).
Collaborators in this project assume that the intricate connection of markers of additivity, continuation, repetition, restitution, contrastive assertion and negation with the language specific ways of unfolding information in connected discourse presents a challenge for L1 learning children as well as for – even advanced – adult second language learners. It is therefore planned to carry out a set of comparative studies investigating productions of these learner populations based on the same stimulus. With respect to L2 acquisition it seems particularly fruitful to chose source and target language combinations based on the type of cross-linguistic differences in language specific preferences mentioned above.
Domain II: Time
With respect to first language acquisition, research in the temporal domain will pursue previous research concerning the effects of predicate types and discourse context from early on, focusing on semantic and pragmatic determinants of children’s uses in very early productions (Bassano, Dressler). In addition, comparisons will involve studying the emergence of these markings in different acquisition situations: monolingual children and children acquiring two languages simultaneously. Integration with research concerning adult learners of L2 will be most crucial in determining cognitive vs. language-specific determinants of acquisition (Demagny, Leclercq).
In L2 adult performance, previous work on narrative construction in English and German (Carroll & Lambert, 2003, 2006 ; von Stutterheim & Carroll 2005) has shown that English speakers use both the simple form (steals) and the progressive (is stealing) in the main structure of retellings. German speakers do not have the option of using a grammaticalised form for ongoing events, and rely exclusively on the expression of bounded events in the main structure. Describing ongoing events as a core part of the narrative in English entails the use of a temporal frame with a deictic anchor (now) and a narrator perspective (now you see, then you see NP Ving). In contrast, Germans organise narratives around bounded events, with anaphoric temporal relations giving temporal shift. Adverbs such as dann (then) are frequent and occur in topic position, with the expression referring to the protagonist occuring after the finite verb. Hence it is no coincidence that very advanced L2 speakers of English are unsuccessful in uncovering how the relevant time-event relations are organised. Comparative work on the associations of form and discourse functions in adult learner and native speakers of languages with a very different linguistic organisation of the temporal domain is planned for Japanese (Tomita), Semitic languages (Ghantous, Bouhaous, Esssam), Slavic languages (Kapral, Shahonenko) and Romance languages (Leclercq, Natale, Castro Alonso).
Domain III: Space
In the study of spatial expression during first language acquisition, previous strong findings obtained among monolingual children will be expanded. As summarized above, previous results showed remarkable cross-linguistic differences in the diversity and density of predicate and structure types at all ages (from two years on). Comparisons with other types of learners and with other languages will help disentangle different factors involved in the acquisition process. It is hypothesized that such comparisons will provide emprical support to generalize the claim that language-specific properties affect children’s attentional/cognitive organization (also see “emergent themes” below).
(1) posture verbs, such as sit, lie, stand, hang, squat, kneel, lean;
(2) caused posture or position verbs, such as lay, wrapped, pinned, stuck, glued, etc.;
(3) other expressions referring to manner of location, e.g., a stack of, an array of, etc.
Results showed large cross-linguistic differences in uses of these different devices, with a massive reliance on posture and position verbs in Germanic languages as compared to rare uses in French. In contrast, manner of attachment is more frequently expressed in French than in Germanic languages, a result also found with children (Hickmann, 2006; Hickmann & Hendriks, 2006). Preliminary analyses further suggest a more fundamental opposition in the topic of the descriptions, viz. the tendency to talk about people and actions (French) rather than about entities and locations (English). This result will be further evaluated in the present project. In addition, native productions will be compared with those of second language learners acquiring a Germanic language from a Romance native language, and vice-versa.
New data will be collected (Lemmens, Tutton) with two groups of advanced adult L2 speakers of these languages: English L2 by French L1 speakers (V-language to S-language) and French L2 by English L2 speakers (S-language to V-language). Data will be elicited during experiments in which subjects will be asked to give a description of a number of pictures. To maximise comparability of the L1 and L1 data, the same experimental set-up will be used as in earlier experiments (cf. Lemmens et al. 2007). It is also planned to analyse the co-verbal gestures, comparing L2 speakers to L1 speakers, in order to evaluate the language cross-over on both the lexical and gestural domain (Lemmens, Gullberg, Yoshioko, see below).
B. Aims of comparisons across learners in L1/L2
The aim here is to compare the processes and outcomes of child and adult acquisition. A more general question that cuts across the three research domains concerns “end states” in first and second language acquisition: whereas ‘normal’ children end up mastering the language of the social environment, adults are notoriously less successful.
B1 Sequences and explanations
Previous longitudinal work on both children and adults has thrown to light acquisitional sequences which are superficially similar. It is then necessary, when faced with such sequences, to determine whether the similarity can be attributed to the same causes, or not. In our own work, we have found both answers, illustrated by the following examples.
(i) The developing expression of finiteness shows similarities
In previous work involving the
(ii) The developing expression of additive relations shows underlying differences
In previous work involving the Nijmegen, Paris and Lille teams (Benazzo, Dimroth, Perdue & Watorek 2004)) we compared children’s and adults’ acquisition and use of additive particles as a cohesive device in German and French discourse, by having subjects relate a cartoon book specifically designed to elicit additive relations. Both languages have a particle aussi, auch which can be applied unrestrictedly to entities, times, places and events, as well as more specific particles for entities - encore (un), noch (ein) – events – encore une fois/de nouveau, (schon) wieder/nochmal – and times – encore/toujours, (immer) noch. It was found that both types of learner acquire these particles in the same order:
addition of entities/places > addition of discrete events > addition of temporal intervals
However, this sequence occurs for different reasons. Four-year-old children first use aussi/auch as a way of relating extra-linguistic entities/places (pointing to the pictures) and these particles are only used later as a cohesive device in discourse. Furthermore, the temporal use of additive particles is a late development. The adults, on the other hand, master the whole range of additive relations in their L1 (Polish) but can only apply this knowledge gradually as their L2 linguistic means allow: - the addition of temporal intervals requires the morphology to express imperfectivity, for example. The same developmental sequence is thus determined: - by the child’s progressive mastery of discourse coherence/cohesion; - by the adult’s progressive development of a linguistic repertoire.
The emerging picture is then that structural difficulties form-function association may pose similar problems to all learners of a specific language, differences in cognitive development may underlie even superficially similar sequences.
B2. Outcomes: ‘end-states’
The first important attempt to explain adult « fossilisation » was Lenneberg’s (1967) hypothesis of a ‘critical period’ for language acquisition situated roughly at puberty. The hypothesis has been developed in much subsequent work. Whereas global cognitive explanations may account for the relative lack of success of the adult learner, they do not account for the variable success of such learners. Much work (for a recent overview, see Singleton & Ryan, 2004) has therefore been devoted to explaining such variability. Results show, firstly, that native-level mastery of the L2 is not beyond the capacity of late learners and, secondly, that other factors than cognitive development need to be taken into account, in particular, the learner’s communicative needs and motivations.
The approach adopted in this work and the results obtained have allowed us to analyse in detail – and in contrast to much other work – the whole acquisition process, from the very beginning (Gullberg, Roberts, Dimroth, 2005) right up to different ‘end-states’. Two points of potential fossilisation along the path to mastery of the target language have received particular attention: the « basic variety » (Klein & Perdue 1997), and « near-native competence ». The former represents a self-sufficient, if limited, lexically-based system of communication that provides insights into a prototypical (Watorek 1996) procedure for organising information in connected speech available to any adult speaker. The latter represents mastery of the sentence grammar of the target language but not the way this knowledge is applied by native speakers in connected discourse (Carroll & Lambert 2003; von Stutterheim & Lambert 2005). These two points are particularly relevant for the research areas of the present proposal, and we give two examples as an illustration.
Basic variety speakers
In previous work involving the
But whereas children very quickly pass through this ‘basic’ stage on the way to mastery of the TL verbal system, adults do not. Some remain at this basic stage, whilst others go on to a (variable) mastery of some or all of the verbal morphology. One attempt to explain this variation is to see both what communicative possibilities the basic variety offers, and what is communicatively not possible. In the transition towards finite utterance organisation, it has been shown (Starren 2001, Benazzo 2003) that there is a complex, language-specific interplay between the development of tense-marking and adverb use. Learners of Dutch use auxiliaries is, heeft, was in different utterance-internal positions to distinguish tense and grammatical aspect marking, whereas learners if French mark tense by morphology and grammatical aspect by adverbs (e.g., déjà ‘already’ for completed aspect) due to the multifunctionality of French verbal inflexional morphology. In both cases however, tense and aspect are expressed by separate means. Some learners distinguish the temporal and aspectual components, expressing them in an analytic way before a sub-set of these learners package both values in one synthetic (target) form. This observation strongly relativises the generality of the “aspect hypothesis” discussed above, and provides a communicative explanation for the adult’s non-acquisition of target morphology.
Near-native L2 speakers
The acquisition of a grammatical form and its associated meaning by an adult learner does not necessarily lead to its correct use in discourse organisation. A well studied-example of this generalisation is the use of the English progressive marker by advanced learners with German and French as their native language. It has been shown (Carroll & Lambert 2003, von Stutterheim & Lambert 2005) that such learners do not become aware that ongoing events, marked by the progressive, form a core part of the foreground of English narratives, serving to shift the story-line. The discourse of such learners, while grammatically error-free, is nevertheless clearly non-nativelike. The general conclusion is that the mapping of correct target forms to the discourse constraints they obey is a final, often insurmountable, stumbling block for the adult learner.
These two illustrative examples underlie a research programme that is in its infancy, as explanation presupposes that the relevant acquisitional descriptions are available, and, in the second case, that the target language’s discourse functions are understood. This is far from being the case.
C. New themes expected to emerge from the project
C1. An integrative approach to information structure and temporal/spatial relations
There has been much previous work, in both first and second language acquisition, on the expression of space and time. There has been less systematic work on the interaction of linguistic and informational structure. What we propose to study, in addition to these specific areas is the interaction between the linguistic tasks involved in these areas in language production. We have already seen that a major obstacle for the advanced adult learner who has managed to acquire a new form-meaning association is its discourse function. Adult German-speakers acquiring the English progressive be-ing nevertheless do not master the discourse functions the morpheme assumes in the chronology of native narrative discourse. We give four short examples in order further to illustrate the interrelation of these phenomena from an acquisitional perspective:
· Focus and topic marking. Explicit topicalisation or focus marking emerges when the information structure of the basic utterance is at variance with the discourse-dependent status of one of its constituents (Klein & Perdue 1997), or with the intended scope of some operator, e.g., negation (Becker 2005).
· Finiteness. Finiteness-marking on the (Romance or Germanic) verb, hence person/number agreement, tense and aspect, emerge to allow the learner to express a dissociation between the (Reichenbachian) reference time and event time, or to mark that the inherent temporal properties (aktionsart) of a predicate are at variance with the discourse status of the corresponding utterance (Dietrich, Klein & Noyau 1995; Starren 2001).
· Scope particles and temporal adverbials. The successful integration of these items are dependent on the acquisition of finiteness-marking and other grammatical temporal features (Benazzo 2002, 2003; Dimroth, 2002)
expressions and utterance structure. In previous work it has been clearly shown that there is a relationship
between the expressive possibilities for time and space in a language, and the
structure of utterances in context. For space, Carroll et al. (2000)
s the rich anaphoric paradigm of German adverbs in da- (daran,
davorne, etc.) and German’s V2 constraint to the way native speakers
conceptualise spatial descriptions around the notion of SPACE with anaphoric
chains based on da-, , and contrasts this with English speakers whose
language lacks such a rich paradigm and has no V2 constraint. Such speakers
create text cohesion around the notion of PLACE, with anaphoric chains based on
pronouns. As a consequence, the main structure utterances of descriptions in
each language are organised in a systematically different way.
In addition, as we have seen above, the expression of static and dynamic spatial relations is highly constrained by typological factors that govern where relevant information will be distributed within the utterance. In particular, depending on the language the learner is acquiring, the information to be “packaged” by lexical and/or grammatical means varies considerably. As a result, learners’ utterance structure is heavily influenced by the particular lexicalization and grammaticalization patterns displayed by target/source language(s).
· The expression of temporal and spatial reference. Despite a close linguistic interrelation between the temporal and spatial domains, the question of how these two domains may related during language acquisition has scarcely been addressed. Thus, the notion of “boundedness” or “telicity” in the temporal domain is closely related to the distinction between motion events that involve a change of location (bounded) or not (unbounded). Some of our research focusing on learners’ expression of motion incidentally clearly indicates that the temporal-aspectual markers differ as a function of different types of motion events.
Such interrelations are at the heart of the proposed research. We aim to find out more about processes of acquisition that go beyond the acquisition of specific forms with their systemic meaning, towards (a) an analysis of the interaction of the forms with others and (b) an explanation of their underlying function in structuring information in coherent texts. Furthermore, the psycholinguistic investigation of the use of different linguistic sub-systems should allow us to uncover typological distinctions which are invisible to the static contrastive analysis of linguistic systems. An obvious example from the temporal domain is the distribution of the French ‘equivalent’ of the progressive morpheme be+ing, namely en train de (Leclercq 2005).
C2. Language learning and (re)conceptualisaton
A strong emergent question that arises in this project on language acquisition is the extent to which language learning implies some cognitive (re)structuring. This question is perhaps the most difficult challenge to tackle in future research on language acquisition. Although the project cannot directly address this question empirically, one of its aims is to provide further emergent research directions in this respect.
With respect to children’s first language acquisition, one of the most debated and controversial question in current psycholinguistic research is the extent to which learning a particular language results in a different cognitive organization (cf. Slobin’s 2006 “thinking for speaking” hypothesis). This strong claim requires examining language productions in many languages and in different types of learners, as planned in this project. However, in order to avoid the circularity that is inherent in studying language effects by examining only language productions, it will require future research involving measures of cognitive organization that wil allow us to get a better access to speakers’s internal representation and that do not depend entirely on language use itself. It is hoped that the project will provide a strong basis for setting up research directions in the future. For example, one promising direction is to study co-verbal gestures is integrated in some of the research projects described above.
Another way to get at representations is to study categorization behaviours. For example, with respect to adult language learners, an important question concerns semantic-conceptual development. Advanced L2 learners with good formal accuracy diverge from native speakers in their use of spatial prepositions (e.g. Becker & Carroll, 1997; Carroll, 1993; Ijaz, 1986), verbs of voluntary motion (e.g. Cadierno, 2004; Hohenstein, Eisenberg, & Naigles, 2006; Negueruela, Lantolf, Rehn, Jordan, & Gelabert, 2004), and more global spatial frames of reference or perspectives applied to spatial events (Carroll, 1997; Carroll, Murcia-Serra, Watorek, & Bendiscoli, 2000), to mention but a few spatial domains. They also differ from monolingual native speakers when categorising shapes (Cook, Bassetti, Kasai, Sasaki, & Takahashi, 2006), in their use of number and numeric systems, causal verbs, etc. (e.g. Boroditsky, 2001; Boroditsky, Schmidt, & Phillips, 2003; Wolff & Ventura, 2003). Such effects may be taken as evidence that learners' semantic-conceptual representations differ from the monolingual native representations. Yet, despite these studies, we know remarkably little about how and when learners come to reorganise their representations, what type of shifts may occur depending on the relationship between categories across languages, especially in actual usage beyond the single word level. Our work in domain thus aims to improve our understanding of mechanisms of semantic-conceptual development in production and comprehension.
We feel that themes C1 and C2 are important ones, and indicate, although
in a preliminary way, the direction future research on language acquisition
will take. Furthermore, they cut across the three main research domains
proposed in this document. We request funding for one Post-Doc position, and one
PhD position for each theme: one Post-Doc and PhD student will work in
D. Publications and conferences
The results from this joint French-German project will be presented in international conferences and published in specialized journals recognized by linguists and psycholinguists in the field of language acquisition. In addition, we have planned an international conference at the end of the project (2010), as well as the publication of a collected volume based on the presentations at this conference. It is also expected that the Post-Docs will publish a major article on themes C1 and C2 described above.
E. Social relevance and potential applications of the findings
The project will provide varied types of information concerning language acquisition that are relevant to two types of applications, both of which are of high interest in the context of present social needs. The first one concerns the diagnosis and remediation of language disorders, particularly in children who present various linguistic deficits during the acquisition of their native language and/or of other languages, for example in migrant situations. The second one concerns language teaching in the classroom, addressed either to adults or to children acquiring a second language. The proposed research should provide crucial information concerning developmental sequences as a function of cognitive level (children), of linguistic competence in L1 (children) or in L2 (children and adults), and of (source and target) languages, that can be directly applied towards the planning of adequate teaching programs.
3.2. Program, methodology, calendar
shown in Table 3, two participants will help coordinate details concerning each
domain (one in
Overall coordination of the project
Research within each domain and topic will be carried out across teams in both countries with regular meetings to ensure bilateral collaboration. Most of the proposed research will benefit from considerable data bases stemming from the participants’ previous research. These data bases are summarized in Table 5 (on the following page). These data bases involve a large number of languages and of learners with different acquisition profiles. They consist mostly of two types of language productions (some comprehension data is also available with children): 1) cross-sectional language productions elicited in experimentally controlled situations and 2) longitudinal and cross-sectional spontaneous productions recorded in natural situations. These data bases will be expanded to cover the needs of the current proposal, but the availability of such considerable daba bases at the start of the project in 2008 constitute a strong starting point for this European co-operation.
Since it is one of the aims of the proposed project to put some of the relevant hypotheses that have evolved from the co-operation in the past to more controlled tests, in addition to the existing battery of elicited production tasks (see table 5), a range of methodologies will be employed including experimental techniques, and gesture analysis.