• Aucun résultat trouvé

Dynamic Assessment of receptive vocabulary and phonology of preschool children with German as a second language

N/A
N/A
Protected

Academic year: 2022

Partager "Dynamic Assessment of receptive vocabulary and phonology of preschool children with German as a second language"

Copied!
171
0
0

Texte intégral

(1)

Thesis

Reference

Dynamic Assessment of receptive vocabulary and phonology of preschool children with German as a second language

MARAGKAKI, Ilektra

Abstract

Despite the growing number of young bilinguals entering the school system in Switzerland, there is lack of appropriate speech and language assessment instruments. In this study we aimed to adapt and validate a promising Dynamic Assessment (DA) screening protocole (DAPPLE) for preschoolers with German as an additional language. Our main question was whether our adapted DA of vocabulary and phonology would better differentiate the skills of two groups of sequential bilinguals, with and without suspicion of language and/ or phonological disorder, compared to static tests. Furthermore, we attempted to validate this DA by exploring its associations with alternative external measures that have been validated with bilingual populations. Overall, our results confirmed our initial hypothesis: the DA generally differentiated more accurately the vocabulary and phonological skills of both groups compared to the used static tests. Also, as expected, most external measures were positively associated with our DA. Nonetheless, the findings of this small-scale, exploratory study, need to be further validated and replicated with larger [...]

MARAGKAKI, Ilektra. Dynamic Assessment of receptive vocabulary and phonology of preschool children with German as a second language. Thèse de doctorat : Univ.

Genève, 2021, no. FPSE 781

DOI : 10.13097/archive-ouverte/unige:154220 URN : urn:nbn:ch:unige-1542206

Available at:

http://archive-ouverte.unige.ch/unige:154220

Disclaimer: layout of this document may differ from the published version.

1 / 1

(2)

Section des Sciences de l’éducation Sous la direction de Marco G.P. Hessels et Erich Hartmann

Dynamic Assessment of receptive vocabulary and phonology of preschool children with German as a second language

THESE

Présentée à la

Faculté de psychologie et des sciences de l’éducation de l’Université de Genève

pour obtenir le grade de Docteur en Sciences de l’Education par

Ilektra MARAGKAKI de

Athènes, Grèce

Thèse No 781

GENÈVE Juillet 2021

Numéro d’étudiante: 17- 343 - 039

Jury :

Dr. Hansjörg Abegglen, Université de Genève Prof. Céline Chatenoud, Université de Genève Dr. Hélène Delage, Université de Genève

Prof. Erich Hartmann, Université de Fribourg Prof. Marco Hessels, Université de Genève Prof. Emmanuel Sander, Université de Genève

(3)

Declaration by author

I hereby declare that the conditions for the realisation of the dissertation entitled

Dynamic Assessment of receptive vocabulary and phonology of preschool children with German as a second language

respect the charter of ethics and deontology of the University of Geneva. I am the author of this text and certify that any statement or idea contained in it which is not the result of my personal reflection is attributed to its source and clearly referenced. Any passage copied from another source is placed in quotation marks.

(4)

i

This study is dedicated to all the children who grow up bilingually, including my own.

(5)

ii

Acknowledgements

There are many people who have contributed in one way or another to the realisation of this project.

Dr. Nathalie Hasson was the first one who introduced me to the concept of Dynamic Assessment at City University in London and who encouraged me to further pursue my studies at a doctoral level in Switzerland, speaking highly of Professor Marco Hessels.

I cannot thank enough Professor Hessels for his willingness to supervise my project and the opportunity he offered me to be part of the MEDASI group at the University of Geneva. His contribution has been immense either directly through his scientific expertise on the domain of Dynamic Assessment but, also, indirectly by encouraging me and gently helping me discover my own potential to conduct research.

I want to equally thank my co-supervisor, Professor Erich Hartmann, who has contributed enormously throughout his project. This support was both mental, through his knowledgeable comments on the domain of linguistic and phonological research but, also, practical; without his help the realisation of this project would not have been possible.

A big thank you to Professor Emmanuel Sander and to Dr. Hansjörg Abegglen for being part of the thesis committee!

I also would like to express my gratitude to Professor Céline Chatenoud as well as to Dr. Hélène Delage for accepting to be part of the jury of this thesis. Thank you!

To the speech therapy students from University of Fribourg, Nora, Linda, and Judith, thank you for helping so decisively with the administration of the tests.

Many thanks to the Association of Swiss-German Speech and Language Therapists (DLV) for their financial support for this project.

I, also, feel the need to thank all the members of the MEDASI group for all the useful

discussions and interactions that helped me promote my scientific thinking. Especially, I want to thank Christine Hessels-Schlatter for all the useful discussions at the Atelier

d’Apprentissage.

I want to express my gratitude to all the children, the families, as well as the teachers and speech therapists from Cantons Fribourg and Bern, who kindly accepted to participate in and support this research. Without them, this project would not have been possible.

I would like to thank all my multilingual students at the International School of Geneva. They have really inspired me and showed me first-hand the need for accurate and economical evaluations in Speech Therapy and how Dynamic Assessment might enable this.

Lastly and most importantly, I want to thank my loved ones. My loving partner, Panos, for all his useful comments, support, and patience for whenever I was not available; my parents, Machi and Michalis, for all their love and support; and finally, the star of my life, Aris, for inspiring me and for being patient for all the times “Mum needed to write her book”.

(6)

iii

Table of Contents

1.Introduction – The Swiss Context ... 1

1.1 Theoretical framework ... 2

1.1.2 Theoretical considerations of language acquisition and implications for bilingual development ... 2

1.1.3 Bilingual language development ... 4

1.1.3.1 Definition... 4

1.1.3.2 Types of bilingualism ... 4

1.1.4 Factors affecting typical sequential bilingual development ... 5

1.1.4.1 Individual variability and uneven ability ... 5

1.1.4.2. Age ... 5

1.1.4.3 Exposure to linguistic input ... 6

1.1.5 On typical bilingual language acquisition ... 7

1.1.6 Linking monolingual language acquisition to bilingual language acquisition ... 8

1.1.6.1 Phonological acquisition of bilinguals ... 9

1.1.6.2 Semantic acquisition of bilinguals ... 10

1.1.6.3 Syntax-Morphology of bilinguals... 15

1.2. Language impairment in monolinguals ... 17

1.2.1. Definitions and causes of Language Impairment ... 17

1.2.2 Developmental course and characteristics of LI... 19

1.2.3 Theoretical accounts of LI ... 22

1.3. Language impairment in German monolinguals ... 25

1.3.1 Grammatical disorders in German ... 25

1.3.2 Phonological and phonetic disorders in German ... 26

1.3.3 Semantic-lexical disorders in German ... 26

1.4. Language Impairment in bilinguals ... 27

1.4.1 Alternative terminology ... 27

1.4.2 Linking theoretical accounts of LI to bilingual acquisition ... 27

1.4.3 A brief description of the data regarding the profiles of bilinguals with LI per linguistic area ……….28

1.4.3.1 Speech sound disorders in bilingual children ... 28

1.4.3.2 Semantic deficiencies in bilingual children ... 29

1.4.3.3 Sentence structure weaknesses in bilingual children ... 30

1.4.3.4 General long-term language growth ... 30

1.5 Bilingualism and cognition ... 30

1.6. Non-verbal reasoning tasks and Language Impairment in CLD children: Traditional testing ... 33

1.6.1 Non-verbal reasoning tasks ... 34

(7)

iv

1.7 Measuring Language Impairment in bilinguals ... 37

1.7.1 Aims of Speech and Language Assessment ... 37

1.7.2 A framework of most frequently used assessments with bilinguals ... 38

1.7.3 Recommended practices regarding assessment of Children with CLD... 39

1.8 Issues regarding speech and language assessment of bilinguals: the case of German-speaking cantons of Switzerland ... 40

1.8.1 Assessment process of bilingual children in Switzerland & recommended practices ... 40

1.8.2 Application of recommended practices in Switzerland: Challenges ... 41

1.8.3 The Swiss-German context: most common instruments used with bilinguals; psychometric properties and general issues. ... 43

1.9 Issues regarding speech and language assessment of bilinguals: the case of standardized -static- assessments ... 44

1.10 Alternative assessment methodologies ... 45

1.10.1 Nonword repetition (NWR) as an evaluation method to identify LI in monolinguals ... 45

1.10.1.2 NWR and the assessment of bilinguals ... 47

1.10.2 Parental Questionnaires ... 47

1.10.3 Dynamic Assessment of language ... 49

1.10.3.1 DA of Phonology... 50

1.10.3.2 DA of Vocabulary ... 51

1.11 Research goals and questions of current Study ... 53

1.11.1 Overarching goal ... 53

1.11.2 Research question of pilot study... 53

1.11.3 The pilot study/ adaptation of the DA of Vocabulary section of the DAPPLE ... 54

1.11.4 Research goals and questions of main study ... 55

Research question & Hypothesis 1: ... 55

Research question & Hypothesis 2. ... 55

Research questions 3-6: ... 56

Research question & Hypothesis 3 ... 56

Research questions 4a & 4b: ... 57

Hypothesis 4: ... 57

Research question & Hypothesis 5 ... 58

Research question & Hypothesis 6: ... 58

2. Methodology ... 60

2.1 Design ... 60

2.2 Participants ... 60

2.2.1 Recruitment ... 60

2.2.2 Description ... 61

2.2.2.1 Age and gender of participants ... 62

(8)

v

2.2.2.2 Language characteristics ... 62

2.2.2.3 Socioeconomic status (family education): ... 63

2.3 Μaterials and procedure ... 64

2.3.1 Nonverbal reasoning ... 64

2.3.2 Vocabulary ... 66

2.3.3 Phonology……….73

2.3.3.1 A merged format ... 73

2.3.3.2 The original phonology task (DAPPLE study)... 73

2.3.3.3 Adaptation in German ... 74

2.3.3.4 Adapted version- Phonological static test ... 75

2.3.3.5 Characteristics of the selected phonological items ... 75

2.3.3.6 Phonology procedure (static-dynamic)) ... 77

2.3.4 Receptive language and morphology (grammar) ... 78

2.3.4.1 A brief description and rationale of each selected Lise-DaZ subtest ... 79

2.3.5 Non-word repetition test ... 83

2.3.6 Parental questionnaire ... 84

2.4 Administration of tests -Training of examiners ... 85

3. Quantitative Results ... 87

3.1 Results concerning Hypothesis 1 ... 87

3.1.1 Vocabulary measures ... 87

3.1.2. Phonology Measures ... 89

3.1.3 General conclusion ... 92

3.2 Results regarding Hypothesis 2 ... 92

3.3 Results regarding Hypotheses 3-5 ... 92

3.3.1 Results regarding Hypothesis 3 ... 96

3.3.2 Results regarding Research questions 4 (a & b) and Hypothesis 4 (Lise-DaZ) ... 95

3.3.3 Results regarding Hypothesis 5 ... 99

3.4 Results regarding Hypothesis 6 (AldeQ questionnaire and NWR task) ... 101

4. Qualitative differences between groups... 105

4.1 Phonology ... 105

4.2 Differences between the two groups on Semantic and Morphological ability based on performance on the Lise-DaZ subtests………..……….107

5. Discussion ... 110

5.1 Study overview: Context and main findings ... 110

5.2 Participants and recruitment process ... 112

5.3 Issues concerning the specific results. ... 113

5.3.1 The differentiating power of Dynamic measures of this project (Hypothesis 1) ... 113

5.3.2 Nonverbal reasoning skills (Hypothesis 2) ... 115

(9)

vi 5.3.3 The relationship between PDSS static vocabulary tests and Dynamic Vocabulary performance

(Hypothesis 3)………..………...……….116

5.3.4 The differentiating power of Lise-DaZ (Research questions 4a, 4b) ... 117

5.3.5 Dynamic vocabulary Scores and Lise-DaZ subtests Comprehension of verbs and Production of modal verbs (Hypothesis 4)... 118

5.3.6 Dynamic vocabulary Scores and Lise-DaZ subtests Conjunctions, Sentence Assembly and Prepositions (Hypothesis 5). ... 120

5.3.7 AldeQ questionnaire and NWR tasks (Research question and Hypothesis 6) ... 120

5.4 Qualitative observations ... 123

5.4.1 Differences between the two groups on Phonology ... 123

5.4.2 Differences between the two groups on Semantic and Morphological ability based on performance on the Lise-DaZ subtests ... 123

6. Conclusions, implications for clinical practise, limitations, and future directions ... 125

References ... 128

Appendices ... 145

Appendix 1. Letters of authorisation of research from Cantons Bern and Fribourg ... 146

Appendix 2. Invitation of participation to research aimed at speech therapists ... 148

Appendix 3. Invitation to participation to research (Schools) ... .149

Appendix 4. Letter of consent for parents ... ...151

Appendix 5. Spearman correlations (whole group) among dynamic vocabulary tasks and Lise-DaZ Comprehension and modal verbs subtests ... …153

(10)

vii

List of Tables

Table 1. Summary of the recommended “care pathway” for CLD children in

Switzerland (translated from original)...41

Table 2. Age and gender of participants per group………...63

Table 3. Frequency of each L1 per group/ total...64

Table 4. Educational level per group………...65

Table 5. Stimuli of the static test of vocabulary ………...70

Table 6. Age of acquisition of German phonemes...78

Table 7. List of words of the Phonological task...79

Table 8. Acquisition patterns of sentence development in German………...……….…. 84

Table 9. Number of syllable and position of the Mottier Test………...86

Table 10. Means (SDs), ANOVAs, and effect sizes of performance between the two groups on vocabulary measures……….…..90

Table 11. Means (SDs), ANOVAs, and effect sizes of both groups’ performance on phonology measures………..92

Table 12.Means (SDs), ANOVAs, and effect sizes of both groups’ outcomes on CPM………..95

Table 13. Spearman rho correlations between static vocabulary test and Dynamic measures………...96

Table 14. Means (SDs), ANOVAs, and effect sizes of both groups’ performance on Lise-DaZ Subtests……….…………..97

Table 15. Spearman’s rho correlations (between groups) among dynamic vocabulary and LD Comprehension and Modal Verbs measures……….101

Table 16. Spearman’s rho correlation among Dynamic measures Vocabulary and LD (expressive) Subtests Sentence Assembly, Prepositions and Conjunctions (between group)……...……102

Table 17. Means (SDs), ANOVAs, and effect sizes of both groups’ performance on NWR and AldeQ………...……….104

Table 18. Spearman’s correlations among Dynamic measures of Vocabulary & Phonology, AldeQ and NWR tasks (whole group)………..………..105

Table 19. Between-groups’ spearman’s correlations among Dynamic measures of Vocabulary, AldeQ, & NWR tasks……….…….106

Table 20. Between-groups’ spearman’s correlations among Dynamic measures of Phonology, AldeQ & NWR tasks………..……….106

Table 21. Number (and percentage) of children per group and per phonological process………...…108

Table 22. Percentages of “red-flag” inconsistency rates per group………..109

(11)

viii

List of Figures

Figure 1. Example of two DNVR tasks……….………….66

Figure 2. Example of scoring form of the static vocabulary test………...69

Figure 3. Presentation of materials of the Dynamic Vocabulary phase……….70

Figure 4. Scoring form of the Dynamic mediation phase of vocabulary………...……….71

Figure 5. Scoring form of the Immediate Expressive Recall of vocabulary………...………72

Figure 6. Scoring form of the Expressive Retention phase of vocabulary……….72

Figure 7. Scoring form of the Receptive Retention phase of vocabulary………...73

Figure 8. Bar graph of “red-flag” inconsistency rates* per group………...…….106

Figure 9. Frequency of raw scores of each group on the subscale Comprehension of verbs……...…108

(12)

ix

Résumé en français

Le contexte de l’ étude

Près de 25% de la population résidant en Suisse, âgée de 0 à 19 ans, est d’une nationalité étrangère (Bundesamt für Statistik, 2014). Il s'avère que les compétences linguistiques des enfants d'origine étrangère sont essentielles pour leur réussite scolaire. Selon les résultats du programme international pour l’évaluation des élèves (PISA) - qui comparait les compétences en lecture des enfants migrants à la fin de la scolarité obligatoire avec les compétences des élèves migrants suisses - étaient parmi les plus faibles de tous les pays inclus dans l'étude (BFS & EDK, 2002). Bien que la politique de soutien de l'acquisition langagière des enfants provenant d’un contexte culturel ou linguistique étranger (« culturally or linguistically diverse » - CLD) ait considérablement évolué depuis lors, il apparaît encore aujourd'hui que les circonstances et les perspectives éducatives de ces enfants sont loin d'être idéales. Par exemple, dans le canton de Zurich, la représentation des élèves d'origine étrangère dans le système scolaire atteint jusqu'à 30% de la population scolaire des écoles spécialisées. Ce contexte, ainsi que la migration mondiale en constante augmentation, soulignent l'importance du bilinguisme dans la recherche actuelle sur l'éducation et la linguistique en Suisse. Plus précisément, notre étude est axée sur l'évaluation linguistique des enfants d’âge préscolaire ayant l'allemand comme deuxième langue dans le contexte de la logopédie.

L’évaluation logopédique des enfants bilingues ou « CLD »

L'objectif primordial de toute évaluation est de collecter et d'interpréter les informations appropriées pour "prendre des décisions éclairées sur le déroulement de l'action clinique " (Kohnert, 2016). En ce qui concerne les enfants bilingues, l'une des questions les plus cruciales à laquelle l'évaluation doit répondre est de savoir si un enfant présente une différence de langage due à un contexte

d'apprentissage allophone ou un trouble spécifique du langage (« Primary Language Impairment »- PLI).

Les tests statiques

La plupart des tests traditionnels en logopédie sont de nature "statique". Ils évaluent les connaissances préalables des enfants dans un domaine et peuvent mener à des diagnostics incorrects de troubles spécifiques du langage (sous-identification). Par conséquent, les enfants d'âge préscolaire atteints de troubles du langage ne sont souvent pas suffisamment dirigés vers des services spécialisés, car leurs

(13)

x difficultés peuvent être attribuées à des problèmes d'apprentissage d'une seconde langue (Kapatzoglou, Restrepo, & Thompson, 2012). En revanche, comme les bas résultats aux tests statiques peuvent ne pas refléter le véritable potentiel des enfants, cela peut résulter à des références excessives aux

services spéciaux (sur-identification). Ces problèmes sont principalement dus à des biais de contenu et des biais linguistiques. Les tests standardisés se basent sur la comparaison des capacités existantes d'un enfant dans un certain domaine, c'est-à-dire le langage, aux capacités d'enfants du même âge au développement typique. Cependant, l'échantillon normatif de la plupart des tests de langue comprend des enfants monolingues qui, souvent, ont des expériences culturelles et linguistiques très différentes de celles des enfants bilingues. Ce fait compromet sérieusement la validité des résultats de

l'évaluation, qui peuvent être biaisés tant sur le plan linguistique que sur le plan du contenu. Il n'est pas surprenant que les enfants bilingues obtiennent souvent de moins bons résultats à ces tâches.

Malgré ces problèmes, les outils de ce type ne sont pas seulement utilisés de manière routinière pour l'évaluation des monolingues, mais sont aussi couramment employés avec des enfants bilingues. Outre les exemples suisses (Häusermann, 2009), ce fait est également confirmé par diverses études

internationales concernant les pratiques des logopédistes avec cette population (Cesar & Kohler, 2007;

Williams &McLeod, 2012).

Les tests dynamiques

L'évaluation dynamique (ED) est une forme alternative de tester qui inclut une certaine forme

d'enseignement ou de médiation, afin d'estimer la zone de développement proximal d'un enfant et son potentiel d'apprentissage. Bien que l'évaluation dynamique ait été mise en œuvre comme un moyen culturellement plus équitable d'évaluation cognitive depuis plus de 50 ans (par exemple Budoff &

Friedman, 1964 ; Carlson & Wiedl, 1979 ; Campione & Brown, 1987 ; Feuerstein, 1980 ; Hessels, 1997 ; Jensen, Robinson-Zanartu & Jensen, 1992), ce n'est que relativement récemment que les chercheurs en langues ont reconnu son potentiel pour déterminer le potentiel d'apprentissage du langage des enfants (par exemple, Camilleri & Law, 2007 ; Guttierez-Clellen, Brown, Conboy, &

Robinson-Zanartu, 1998 ; Olswang & Bain, 1996 ; Peña, Iglesias & Lidz, 2001).

La performance des enfants aux tâches d'ED dans des domaines tels que l'apprentissage des mots suggèrent que lorsque ces informations sont combinées avec leurs performances post-test à un test statique connexe, cela permet une classification plus précise de leur capacité d'apprentissage par rapport aux tests statiques seuls (Peña, Iglesias, & Lidz, 2001 ; Ukrainetz, Harpel, Wall, & Koyle, 2000). Plus précisément, les résultats de ces études suggèrent qu'en ce qui concerne la validité prédictive, un enfant hautement "modifiable", c'est-à-dire ayant un score élevé à l'ED, qui présente également des gains au test statique entre différents points dans le temps, est plus susceptible d'avoir

(14)

xi des capacités typiques par rapport à un enfant montrant peu de changement à l'ED et aux mesures statiques (Hessels, 1997, 2002). Cette méthode de test peut donc fournir des informations qui ne pourraient pas être obtenues autrement et améliorer la qualité de l'intervention proposée.

En outre, l'ED peut être particulièrement utile lors du dépistage des enfants bilingues ou « CLD », car elle examine la capacité à acquérir de nouvelles compétences plutôt que des connaissances acquises.

Un tel outil d'ED serait administré uniquement en L2 (deuxième langue), sans inclure la première langue de l'enfant, car cela constituerait une "connaissance acquise", contribuant ainsi à une évaluation plus économique. L'ED est l'une des pratiques recommandées au niveau international pour l'évaluation bilingue (Caeser & Kohler, 2007) et a été proposée comme alternative pour l'évaluation bilingue administrée par les logopédistes suisses-allemands (Frigerio-Sayilir, 2013). Un exemple prometteur d'une telle évaluation est un outil de dépistage visant à distinguer la « différence » langagière due au contexte du trouble spécifique du langage chez les enfants bilingues anglophones, le « Dynamic Assessment of Preschoolers' Proficiency in Learning English » (DAPPLE; Hasson, Camilleri, Jones, Smith & Dodd, 2013). Cet instrument destiné aux enfants âgés de 3 à 5 ans comprend un dépistage au niveau phonologique, au niveau du vocabulaire et au niveau de la structure des phrases

(morphologique). La présente étude a été basée sur DAPPLE et, plus concrètement, elle se focalise sur la phonologie et le vocabulaire réceptif.

La présente thèse : Procédure, Questions et Résultats principaux

Le principal objectif de cette étude était de créer une mesure alternative, précise et économique, à travers d'une ED, pour mieux distinguer la « différence » langagière due au contexte du trouble spécifique du langage chez les enfants bilingues (séquentiels) germanophones en Suisse. Notre but était d'adapter et de valider deux sous-tests (Vocabulaire et Phonologie) du DAPPLE avec des enfants d’âge préscolaire, à la suite d’un projet pilote prometteur à petite échelle (Maragkaki & Hessels, 2017). Plus précisément, la structure du sous-test du vocabulaire consistait en une première phase d’une tâche d'identification d'images (phase statique de l’évaluation du vocabulaire réceptif). Elle était suivie d'une courte tâche de médiation basée sur le nombre de mauvaises réponses au test statique. Les trois conditions dynamiques étaient les suivantes : a) le rappel immédiat, b) la rétention expressive et c) la rétention réceptive.

En ce qui concerne, la partie phonologique elle était basée sur les procédures de DEAP (Dodd, Zhu, Crosbie, Holm, & Ozanne, 2002), DAPPLE et PLAKSS (Fox-Boyer, 2014). Le but général est d'identifier les "aspects du système vocal qui nécessitent une évaluation plus approfondie" (DEAP ; Dodd, Zhu, Crosbie, Holm, & Ozanne, 2002, p.3). Pour ce faire, il faut nommer de manière répétée dix images, entrecoupées d'une tâche du contrôle de stimulabilité des sons (capacité à imiter un son sur

(15)

xii demande). L'idée principale est que si un enfant n'est pas capable d'imiter avec succès les sons que 90% des enfants du même groupe d'âge sont capables d'imiter, alors des tests plus détaillés de l'articulation et du fonctionnement oro-motrice doivent être administrés. Au cours de ce processus, il est important d'examiner le type d'erreurs, c'est-à-dire s'ils sont typiques ou atypiques, car c'est dans ce dernier cas que la thérapie sera prioritaire. En outre, un taux d'incohérence (« Incosistency ») pourrait être calculé en comparant les mots qui ont été produits différemment au cours des deux essais. Un taux d'incohérence de plus de 40 % justifie également un contrôle plus approfondi, car une telle

performance pourrait indiquer une dyspraxie verbale ou d'autres difficultés d'articulation.

Questions et résultats

La première est plus importante question de notre recherche était si l’ED du vocabulaire et de phonologie pourrait mieux distinguer les connaissances des jeunes bilingues (séquentiels) germanophones avec et ou sans suspicion de trouble du langage par rapport à des tests statiques connexes.

En ce qui concerne nos résultats, notre objectif initial a été globalement confirmé: L'instrument de vocabulaire dynamique (tâches de médiation et d'expression) a permis de distinguer avec plus de précision les compétences des enfants de deux groupes, par rapport à la tâche statique (PDSS- Compréhension des noms; Kauschke & Siegmüller, 2002). La performance des enfants à la tâche de rétention réceptive a également démontré quelques indications de différenciation, mais cela n'a pas atteint le seuil de signification. En outre, la performance à tous les items de phonologie dynamique, et en particulier au sous-test des mots, a permis de différencier assez bien les compétences des deux groupes; la performance aux instruments statiques a également permis de différencier les deux

groupes, mais elle était associée à des tailles d'effet plus faibles par rapport aux tâches dynamiques. En d'autres termes, la capacité à répéter des sons, le nombre de sons émis correctement à la suite d'un modèle, ainsi que le taux d'incohérence entre les productions du même mot à deux occasions

différentes constituaient des facteurs de différenciation précis par rapport aux compétences des deux groupes. Ces résultats confirment des données existantes dans ce domaine, tant en anglais (Dodd, 2013) qu'en allemand (Fox-Boyer & Strutzke, 2018).

De plus, on a visé à explorer la relation de nos instruments dynamiques avec certains des instruments et procédures statiques déjà validés et, ainsi, à explorer des indications possibles de la validité convergente et critérielle de l'AD (Camilleri & Law, 2007). Les mesures choisies comprennent le test PDSS qui a été utilisé comme prétest pour cette étude, des subtests spécifiques du Lise-DaZ (Schulz &

Tracy, 2011), une version adaptée du questionnaire AldeQ (Paradis et al., 2010), ainsi que le Mottier (1951, as adapted in Krauser, 2010) test de répétions de non-mots.

(16)

xiii En ce qui concerne les résultats, notre étude confirme globalement que les instruments dynamiques adaptés de vocabulaire et de phonologie étaient significativement associés à une série de mesures différentes. L'instrument de vocabulaire dynamique était modérément associé au PDSS statique (compréhension des noms), qui est une mesure établie de la capacité lexicale (bien que non standardisée avec les enfants CLD). Cela confirme, dans une certaine mesure, l'utilité de notre AD comme mesure du potentiel d'apprentissage lexical (Law & Camilleri, 2007). De plus, le fait que l'AD du vocabulaire soit associée à plusieurs sous-tests du test Lise-DaZ, qui a déjà été validé avec des enfants bilingues, comme la tâche de compréhension des verbes, ainsi que la tâche de production de verbes modaux, de prépositions et d'assemblage de phrases, confirme encore l'efficacité de notre AD pour cette population. Bien que dans plusieurs cas, les tâches de Lise-DaZ aient été assez difficiles pour les participants de cette étude, et malgré les quelques enfants du groupe référé (« Caseload ») qui ont pu répondre à certaines des tâches (minimisant ainsi la possibilité d'une corrélation significative), il y avait des différences qualitatives substantielles entre les deux groupes. Par exemple, la différence concernant la capacité des deux groupes à former des phrases (sous-test d'assemblage de phrases) correspond aux types de difficultés auxquelles les enfants avec trouble du langage sont fréquemment confrontés en allemand (Schulz & Tracy, 2011). Enfin, plusieurs de nos mesures de l'AD étaient également modérément associées aux tâches de répétition du NWR (test de Mottier), établissant que ces mesures évaluent des construits similaires, au moins partiellement. Cela était moins évident avec le questionnaire parental (AldeQ), du moins comme le montrent les corrélations entre groupes.

En plus des différents tests phonologiques, de vocabulaire et morphologiques, il était important d’obtenir une estimation des capacités de raisonnement non verbal des enfants. En suivant le paradigme du DAPPLE, nous avons utilisé un test statique, qui dans notre cas était le CPM (Colour Progressive Matrices; Raven, 1998). Cependant, en même temps, en raison des problèmes

méthodologiques qui ont été associés aux tests statiques des enfants CLD, nous avons décidé de mesurer également les performances des enfants de manière dynamique. Une mesure dynamique du raisonnement analogique non verbal, le DMVR, a été spécialement conçue pour cette étude sur la base du LEM (Hessels, 1997) et de l'ARLT (Hessels-Schlatter, 2002). Nos résultats confirment notre hypothèse selon laquelle aucune différence ne serait observée en ce qui concerne la performance des deux groupes au DMVR, contrairement aux petites différences quant à la performance au CPM.

(17)

xiv

Summary

Language competence of children of foreign background has been found to be crucial for their educational success. Nonetheless, the process of assessing such populations is especially challenging for the professionals involved, such as speech and language therapists who are the focus of this study.

This issue is not only evident at a national, i.e., Swiss level, but, also, internationally. One of the most important challenges during assessment of bilingual children is that this process is based on so-called static instruments. Since these assess children's prior knowledge, they can lead to incorrect diagnoses of SLI (identification). As a result, pre-schoolers with language impairments are often under-referred to specialized services because their difficulties may be attributed to second language learning problems (Kapatzoglou, Restrepo, & Thompson, 2012). On the other hand, because low static test scores may not reflect children's true potential, this may result in excessive referrals to special services (over-identification). Due to this reason international organisations such as ASHA recommend that speech therapists use alternative methodologies when assessing culturally and linguistically diverse children. Dynamic Assessment (DA) which includes some mediation during the evaluation process has proved to be such an alternative. It has been found to have better predictive validity and better classification accuracy of the linguistic skills of the children. Also, by measuring children’s potential to learn it provides more clinically useful information as regards the planning of therapeutic

interventions.

The Dynamic Assessment of Preschoolers' Proficiency in Learning English (DAPPLE; Hasson, Camilleri, Jones, Smith & Dodd, 2013) is promising example of such an assessment (screening) aiming to distinguish language "difference" due to the context of specific language impairment in English-speaking bilingual children. This instrument for children ages 3 to 5 years includes screening at the phonological, vocabulary, and sentence structure (morphological) levels.. The main objective of this study was to create an alternative, accurate and cost-effective measure, through a DA, to better distinguish the language "difference" due to the context from language impairment in German- speaking (sequential) bilingual children in Switzerland. Our goal was to adapt and validate two subtests (Vocabulary and Phonology) of the DAPPLE with preschoolers, following a promising small- scale pilot project (Maragkaki & Hessels, 2017).

This thesis starts with a presentation of the main theories of monolingual and bilingual linguistic acquisition, the main characteristics of language impairment in mono- and bilingual children and how this affects assessment. The first section concludes with a presentation of our research questions and

(18)

xv hypothesis. The second section includes a description of the recruitment process and characteristics of the participants, as well as the main instruments that were used during this research, that is our DA of vocabulary and phonology, the static and dynamic instruments of nonverbal reasoning, as well as the instruments that were used as alternative measures to gauge the validity of our DA. At the third and fourth sections contain we present our results, both quantitatively and qualitatively. These are

followed by a discussion of each result considering the presented theoretical background, highlighting both our promising findings but, also, the limitations of our study.

As regards the results of our study, overall, these confirmed our initial hypothesis as the DA generally differentiated better the vocabulary and phonological skills of both groups compared to the static tests that were used. The DA of nonverbal reasoning, also, provided more a more accurate differentiation of both groups compared to the static test. Furthermore, as expected, most -but not all- external measures were positively associated with our DA, which is an initial indication of validity of this DA.

Nonetheless, the findings of this small-scale, exploratory study need to be further validated and replicated with larger samples.

(19)

1

1. Introduction – The Swiss Context

Nearly 25% of the population residing in Switzerland, aged 0-19 years, have foreign citizenship (Bundesamt für Statistik, 2014). Language competence of children of foreign background has been found to be crucial for their educational success. According to the results of the Program for

International Students Assessment (PISA), which compared the reading skills of migrant children at the end of obligatory education, the competencies of Swiss migrant students were among the lowest out of all the countries included in the study (BFS & EDK, 2002). Although the policy regarding the support of language acquisition of Culturally or Linguistically Diverse (CLD) children has

significantly changed since then, it still appears today that the educational circumstances and prospects for these children are far from ideal. In the canton of Zurich, the representation of students of non- Swiss origin in the school system is as high as 30% of the student population in special schools. On the other hand, this percentage drops down to only 25% of the students in primary, secondary, and vocational schools and 10-15% in high-school or university (Bossart, 2011). This type of evidence, along with the ever-growing global migration, highlight the importance of bilingualism in the current educational and linguistic research in Switzerland. Of specific interest for this study is the linguistic evaluation of CLD preschoolers with German as a second language within the context of speech and language therapy.

(20)

2

1.1 Theoretical framework

1.1.2 Theoretical considerations of language acquisition and implications for bilingual development

One of the most fundamental questions in language acquisition theory is whether it is the product of innate mechanisms or environmental factors. There is a long debate regarding the proposed “answers”.

The nativist or formal theory, as expressed by researchers, such as Chomsky (1965) and Pinker (1994), maintains that language development is based on internal, biologically governed, cognitive

mechanisms, regardless of the nature of the language input. These mechanisms which form the language acquisition device are thought to be universal (universal grammar). The similarity of children’ s acquired syntactic structures across languages is frequently referred to as evidence of the nativist approach (Bialystok, 2001).

A further argument in favour of the innateness of the mechanisms of language acquisition is the

“poverty of stimulus” argument based on which children cannot possibly be provided with rich enough linguistic information in their environments to acquire all potential characteristics of their language.

This idea has been expanded and used by researchers, such as Pinker (1994) to explain children’s tendency to “overgeneralise”, i.e., generate inappropriate past tense forms (learnability theory).

These concepts have direct consequences on theories of bilingual language acquisition: Some scholars of bilingual linguistic development have modified the concept of “poverty of stimulus” into that of

“lack of dual stimulus”, which further endorses the biological or innate character of -bilingual- language acquisition; it would be illogical to think that children could possibly be exposed to all the different features (“stimuli”) of both languages within the time that they acquire them (Yip &

Matthews, 2007). On the contrary, the sole logical explanation would appear to be that linguistic structure of both languages are acquired through an internally driven mechanism. In addition, a further argument in favour of “innateness” is the existence of a critical period for language learning, which is central for second language acquisition.

Supporters of formal theories claim that due to the universal character of language acquisition mechanisms, the first language forms a model unto which the second language would be obtained, based on predefined structures that are only superficially affected by the type of input, i.e., language, a child is exposed. In other words, the patterns of acquisition of the two languages would be similar (Bialystok, 2001). Finally, as regards the ever-lasting discussion of the relationship between language and general cognition, formal approaches consider linguistic mechanisms relatively independent of overall cognitive development.

(21)

3 In contrast to the formal models, functional theories place very strong value on the role of the

linguistic input in language acquisition. It is believed that the environment provides ample information for each child, which then “follows” their individual, unique structure–pathway. Within this

theoretical framework, the child’s grammatical structures are formed through experience. If the social environment shapes the way children learn language, it is expected that the patterns of acquisition of bilingual children would be very different from one another. Furthermore, functional models view language as an integral part of general cognitive mechanisms. The role of cognitive mechanisms in shaping linguistic representations has been the ultimate cornerstone of theories, such as construction grammar and connectionist models (Tomasello & Brooks, 1999). A parallel that can be drawn between the latter theories and that of Piaget’s developmental cognitive theory is that linguistic progress is the reflection of the acquisition of specific cognitive “schemas” by children (Ringblom, 2012). In summary, it has been proposed that a useful way of looking at these two distinct approaches is by considering formal theories as means attempting to describe language, whereas functional methodologies as instruments examining communication (Bialystok, 2001).

The need to bridge the form-function dichotomy was addressed by scholars, such as Bruner, who, within his interactionist approach, tried to merge aspects from both theories. For instance, he proposed that the innate language acquisition device is simultaneously complemented by a social, interactive system driving the whole process. In a similar fashion, Locke’s bio-linguistic approach (1993) proposed two, equally important, inter-dependent aspects of language acquisition, namely, a system responsible for morphological development and a social cognition “module”; the latter is supposed to be entirely innate and universal, while the former is thought to be partly shaped by the environment, such as in the case of irregular forms (Locke, 1993).

Apart from the interactionist approach by Bruner (1983), the social interaction theory proposed by Vygotsky (1986), also, emphasizes the importance of the social context for learning language, i.e., that children mainly learn during meaningful interactions and that that their active engagement during this process is crucial. Furthermore, Piaget’s (1929) cognitive theory, also, places great importance on the environment and experience as shaping and activating mechanisms of children’s innate mental cognitive structures (Ringblom, 2012).

Finally, another more recent attempt to explain language acquisition has been suggested by researchers such as Deacon (1997), which is inspired by evolutionary cognition. According to this theory,

language was developed as a form of a general, symbolic system, millions of years ago, which was the result of the evolution of specific parts of the human prefrontal cortex. More specifically, the

development of the said system, enabled the human species to represent complex and abstract systems of meaning and, thus, be able to create highly complex societies and cultures. Focusing on the

development of fundamental conceptual abilities, allows this theory to consider innate cognitive-

(22)

4 conceptual abilities without making it specific to a certain language and thus, again, overcoming the previously discussed dichotomy (Bialystok, 2001).

1.1.3 Bilingual language development 1.1.3.1 Definition

Defining bilingualism is as complex as the phenomenon itself. For instance, while for some researchers bilingualism is synonymous with “functional fluency in both languages” (Bialystok, 2001), for others being bilingual entails a much more complex and unique linguistic configuration that should not be simplified into the idea that bilinguals are just the “sum” of two monolinguals in one (Grosjean, 1989). One way of categorising the multitude of existing definitions is by considering the aspect of assumed proficiency of the bilingual speakers. There is a plethora of definitions of

bilingualism ranging from those who imply that bilinguals should exhibit “native-like proficiency in both languages” (Bloomfield, 1933; Ng, Chin & Wigglesworth, 2007) to less restrictive ones, such as the one proposed by Mackey (2000), namely that bilinguals are termed “those individuals who use two or more languages in any modality, namely speaking, reading or writing” (as cited in Hasson,

Camilleri, Jones, Smith, & Dodd, 2013). For the purposes of this research, we will adopt the latter definition, in line with the DAPPLE study (Hasson, Camilleri, Jones, Smith, & Dodd, 2013).

1.1.3.2 Types of bilingualism

Another important theoretical issue in the study of bilingualism is the type of bilingualism. This notion has been researched in an attempt to answer questions, such as the status of both languages, the relationship between bilingual language acquisition and second language acquisition, as well as the nature of development of the two different linguistic systems of a child.

The main distinction used in the literature of child bilingualism is that between “simultaneous” and

“successive” or “sequential” bilinguals, with each “type” being, frequently, associated with distinct research methodologies. “Simultaneous” bilinguals are termed those children who acquired both languages within the first year of their life (Armon-Lotem, De Jong & Meir, 2015). On the other hand,

“successive” or “sequential” bilingualism refers to children who learned one language earlier than the other, often referred to as L1 (first language) and L2 (second language), respectively. However, it should be mentioned that the exact cut-off between the first and second language is arbitrary as some researchers place it at one month and others at the age of three (Bialystok, 2001; Bloomfield, 2012). It has been argued that there might be no clear categorisation between first and second language

(23)

5 acquisition but, rather,” a continuum where the dominant language functions as first language and the weaker as second” (Yip, 2013).

Parenthetically, dominance shift is considered an essential phenomenon in bilingual language acquisition and is thought to be a dynamic process, which changes throughout the developmental process, depending on the context and affecting the whole or different linguistic subsystems of each language (Ebert & Kohnert, 2016).

Taking the simultaneous-sequential distinction into consideration, the children that participated in this study would be classified, almost exclusively, as sequential bilinguals. However, as an additional language, also, implies contact with a different culture, the present study, also, includes the term Culturally or Linguistically Diverse Children (CLD children), which will be used interchangeably with the word bilingual. It should be mentioned that in many countries L1 is considered a minority

language either due to lack of opportunities to practice it or because of a lower perceived social status concerning the use of L1 (Ebert & Kohnert, 2016); this has, also, been the case with most of the participants of the current study.

1.1.4 Factors affecting typical sequential bilingual development 1.1.4.1 Individual variability and uneven ability

Two fundamental characteristics of typical language development in sequential bilinguals are: a) significant individual variation that is associated with complex environmental, linguistic and personal factors and b) distributed skills between the different sub-systems of both languages leading to varying levels of proficiency in each one of them (McLeod,Verdon, Baker, Ball, Ballard et al., 2017). As Grosjean (1982) points out, one should not expect specific proficiency or learning pattern (at a certain language) of a bilingual child to be the same as a monolingual one because the representational systems of both languages are different.

1.1.4.2 Age

The role of age on bilingual acquisition has been researched in relation to the existence of a critical period for language acquisition, during which a child is supposed to have an increased ease to acquire language. This concept will be further examined in relation to phonological development. The

existence of a critical period implies a certain time frame with a specific end date, after which this

“advantage” rapidly decays.

(24)

6 There is evidence both in favour of the existence of such a timeframe, as well as research challenging this idea. Support for the existence of a critical period for language learning has been provided by studies indicating that the learning profile, the developmental sequences, as well as the proficiency levels that may be attained following exposure to an additional language, differs between child and adult sequential bilinguals (DeKeyser, 2000; Clahsen & Stoessel-Daeschner, 1993). These differences that are taken as evidence for the existence of a period of increased sensitivity for language learning have been attributed to the differentiated functioning of two separate memory systems, namely the declarative and the procedural memory. The former is thought to be responsible for the storage of information while the latter is in charge for computational operations that are active during language acquisition. Based on this neurobiological approach, the two systems are thought to interact during language acquisition of a child’s first language but this interaction later, when one “exits” the critical period, weakens; this results in a situation where adult sequential learners are able to process language (and more specifically grammar) mainly through their declarative memory (Ullman, 2004).

Furthermore, as mentioned earlier, there is evidence stemming from studies across different languages that age influences not only the levels of proficiency but also the developmental sequences of error patterns produced by sequential bilinguals. These studies -discussed in more detail on the chapter about LI in bilinguals- share the formal theoretical underpinning (see Universal Grammar) that language learning by young sequential bilinguals is ruled by universal rules that do not apply on adult L2 learners, which “shapes” the expectation of a discrepancy as regards the types of observed errors between the two groups.

However, not all researchers accept the idea of a critical period; the idea that this sensitivity for learning reaches a maturational peak after which it ends has been questioned by other scholars, such as Flege, Yeni-Komshian & Liu (1999), whereby linguistic proficiency as measured by grammaticality judgments tasks by participants of different age groups depended not only on age but, also, on type of item. In addition, the argument presented earlier (see Universal Grammar approach) that L2 learners of different ages would present with different types of morphological errors and type of acquisition has been challenged by studies, such as the one by Bialystok & Miller (1999) showing that such

differences depended on the similarity of the used morphological structures in the two languages, regardless of the age of onset of the acquisition of the L2.

1.1.4.3 Exposure to linguistic input

One of the main components that has been associated with faster and less disordered language acquisition of monolingual children compared to -sequential-bilinguals is the type and amount of linguistic input. This type of input, also referred to as input frequency, may include the actual number

(25)

7 of utterances children hear in a language, the way these languages are used by the parents in a

bilingual household, as well as the responsiveness of each parent on the child’s use of each language (De Houwer, 2011). It is interesting to keep in mind the interesting point that Age of Onset is not synonymous with higher linguistic input (Armon-Lotem, Walters & Gagarina, 2011). Also, the importance of age of onset has been found to be moderated by phenomena, such as the “timing in L1 acquisition”, i.e., when a specific structure actually is expected to be mastered in monolingual children (Schulz & Grimm, 2019)

Various studies have shown that linguistic exposure affects both the vocabulary size as well as the amount of expressive output of young bilingual children (Meisel, 2008). Furthermore, there is research confirming that both the richness as well as the length of exposure to the second language positively affect the acquisition of the morphology of that language (Orgassa, 2009; Unsworth, 2008).

Nevertheless, the amount and the quality of linguistic input is inevitably influenced by the same socio- economic, as well as linguistic-cognitive factors that lead to individual variability as presented in chapter 1.1.4.

1.1.5 On typical bilingual language acquisition

This chapter will briefly touch on a few general important issues on theories of bilingual language acquisition, starting with an important question that has been thoroughly researched: the existence or not of any interference between the two languages, as well as its nature.

The two main hypotheses-theories that have been proposed in an attempt to explain interference between the two languages: a) The Unitary System Hypothesis, according to which bilingual children start as monolingual, with a shared lexicon, continuing with a shared grammar system, which is eventually differentiated at a later point in development. Within this model, interference may be kept minimal between the two languages if the domains of use are maintained separate (Voltera &

Taeschner, 1978); b) The Separate Development (Systems) Hypothesis is an alternative, more recent, theory that has been supported experimentally by many studies. The supporters of this view argue that children who have been exposed to two different languages from early on form two distinct systems, controlling either the entirety of the linguistic development (Genesee, 1989) or just the

morphosyntactic aspect (De Houwer, 1990). Although these theories have been developed for simultaneous bilinguals, they have been mentioned and applied in several studies focusing on the explanation of the development of the linguistic systems of sequential bilinguals (Bialystok, 2001).

It has been hypothesized that if the two languages of a bilingual did not interact with one another, there would not be any associations between performance on controlled tasks in L1 and L2;

conversely, if there was interference between the two languages then that would imply positive or

(26)

8 negative correlations between L1 and L2 on said tasks (Ebert & Kohnert, 2016). There is research data supporting both suggestions. For example, there has been evidence of positive correlations between lexical comprehension tasks in L1 (Hmong) and lexical production in L2 (English) in young children (Kan & Kohnert, 2008). On the other hand, according to meta-analyses regarding the link between L1 and L2 vocabulary scores of bilingual English-Spanish school-aged children, the observed results (i.e., correlations) were close to zero (Branum-Martin, Mehta, Francis, Foorman, Cirino et al., 2009). Nonetheless, as mentioned before, the most current view appears to be the separate systems hypothesis, which endorses the idea of lack of linguistic transfer between L1 and L2.

A more recent approach that has been proposed to overcome the theoretical dichotomy is the

Interdependent Development Hypothesis. This theory supports the existence of a linguistic system that is neither completely single nor entirely unique; the main idea is that there may be indications of transfer or interaction between specific domains of the two linguistic systems that present particular difficulties, such as in cases of an interface or of specific, restrictive types of overlap as regards two structures in L1 and L2, respectively. More specifically, external interfaces, such as syntax and pragmatics, are thought to be more likely to transfer, whereas internal interfaces, such as syntax- morphology are less susceptible to such cross-linguistic transfer (Hulk & Müller, 2001).

Finally, linking the interaction and transfer issues with the general factors affecting bilingual

acquisition presented earlier, it is important to refer to the studies that link transfer with Age of Onset.

Meir and her colleagues (2017) investigated the role of cross-linguistic transfer between Russian- Hebrew bilingual children and its relation to Age of Onset of exposure to L2. Based on their results all bilingual groups of their study showed interaction effects regardless of Age of Onset. Their conclusion was that transfer depended on the similarity of the investigated grammatical features. For example, while for “difficult” features, such as articles (which are not present in Russian) late age of onset did have a negative effect on L2, for grammatical features that were facilitated by L1 (i.e., that were similar) there was no effect of age of onset.

1.1.6 Linking monolingual language acquisition to bilingual language acquisition

As mentioned earlier there are several theoretical accounts of language development. Bialystok (2001) argues that one should take an integrative approach when it comes to bilingual acquisition, especially with regards to questions such as whether it follows monolingual acquisition or not. The same author, also, maintains that although the formalist approach is the most dominant in second language

acquisition research, it is the functional one that has yielded some of the most interesting findings and implications for bilingualism, as it is more straightforwardly links language learning with cognitive functioning, social context and individual variation. As cited earlier, individual variation has been

(27)

9 pointed out as one of the main characteristics of typical bilingual linguistic development (Ebert &

Kohnert, 2016).

Whenever appropriate, applications of each theoretical paradigm will be given with specific reference to each of the linguistic subsystems presented below.

1.1.6.1 Phonological acquisition of bilinguals

The onset of phonological development occurs in the first weeks of an infant’s life and is marked by vocalisations and playful sounds, followed by simple patterns of vowel-consonant combinations. In the past, it was argued that this phase is differentiated from the –later- stage of linguistic acquisition but nowadays these two are thought to form parts of a continuum (Jakobson, 1968; Bialystok, 2001).

Infants are born with an innate ability to distinguish not only native but also non-native phonemic contrasts. This ability is “lost” around 10-12 months of age, as their perceptual system is “tuned into”

the phonemic contrasts of their native language (Werker & Tees, 1984). Overall, the sophisticated perceptual abilities of young children appear to be independent of any specific language. This fact, which is considered one of the important arguments in favour of the existence of a “critical” period as regards (second) language acquisition (Bialystok, 2001), is compatible with the formal linguistic theory, whereby language mechanisms are independent of other cognitive devices. In addition, phonological development of sequential bilinguals is thought to be guided by general universal (phonological) patterns of behaviour, such as the phonological processes that are present in all children (Anderson, 2004).

Nonetheless, even though formal theories might explain some characteristics of phonological development, they cannot account for the interactional effects occurring between a child’s different phonological systems, nor do they suffice to explain areas, such as semantic or pragmatic

development.

In general, the phonological systems of bilingual children are differentiated and interact with one another, as proposed by the Interactional Dual Systems Model (Paradis, 2001) in three possible ways:

in a facilitative way (acceleration), in an “inhibiting” way (deceleration) and by incorporating an element from one language to the other (transfer); (Barlow & Fabiano-Smith, 2010). Transfer is of clinical interest and has been linked with phonological issues, such as markedness, i.e., “relative featural complexity”. Across languages, unmarked sounds are acquired earlier than marked ones following a specific hierarchy, which may differ from one language to another. For example, stops are more easily and earlier acquired than fricatives in English (Dinnsen et al., 1990). It has been found that bilinguals tend to follow the same type of hierarchical organization, as well as acquisition of

phonological systems at each of their languages as monolinguals. Moreover, there has been evidence

(28)

10 that different types of interaction (for example, deceleration and transfer) may co-occur at different times during phonological acquisition of both systems. This information is of clinical interest as it shows that therapists may not be able to reach to safe diagnostic conclusions when using only one type of evidence, i.e., accuracy rates versus types of phonological class (Barlow & Fabiano-Smith, 2010).

It is worth mentioning that proof of interaction between different phonological systems stem, also, from studies about children with German as a L2 (Turkish-German sequential bilinguals); (Salgert, Fricke & Wells, (2012). This type of cross-linguistic influence may explain the non-linear nature of bilingual phonological acquisition that is frequently characterized by progressions, regressions, and stagnations (Albrecht, 2017). Furthermore, there is evidence that bilingual children may exhibit different error patterns of phonological processes compared to their monolingual peers. This type of information highlights the importance of avoiding the application of normative data of assessments that have been standardized to monolingual children to bilingual children (Dodd, 2008).

Unfortunately, research of phonological patterns in bilingual children is complicated by the widely diverse forms and functions of different language pairs (Hemsley, 2015). Also, bilingual phonological acquisition may be further convoluted by factors, such as gender, linguistic proficiency, as well as amount of input and output, as has been evidenced by research with Turkish-German children (Albrecht, 2017).

1.1.6.2 Semantic acquisition of bilinguals a. Processes of word learning

Before delving into the different theories of bilingual semantic acquisition, it is necessary to provide a brief framework of the process of word learning.

Younger children, in general, learn new words mainly through two complimentary processes: fast mapping during which a word label is associated with a referent after one to three exposures (Carey, 1978; Alt, Meyers & Figueroa, 2013) and slow mapping, during which representations get more refined through additional exposures to said referent. In bilingual children, there is evidence that this process is highly influenced by the innate patterns of a child’s native language. Also, several studies suggest that mapping occurs through statistical learning (Evans, Saffran, & Robe-Torres, 2009). This process, which is considered to be one of the several mechanisms of implicit learning, appears to occur without any effort or awareness by the child and is affected by factors, such as phonotactic probability, i.e., the probability of certain sounds occurring within certain phonetic contexts (Alt, Meyers & Figueroa, 2013; Vitevich, Luce, Charles-Luce, & Kemmerer, 1997). Phonotactic probability is associated with another lexical characteristic, the so-called neighbourhood density. A word’s

phonological density is determined by how similar phonetically it is with other words, i.e., how

(29)

11 densely it is nested in the neighbourhoods of “phonological similarities” of the mental lexicon (Storkel

& Morissette, 2002). Phonotactic probability, word frequency and neighbourhood density are thought to affect both word perception and production. For infants and younger children, word-referent associations are made through observations and estimations of the reliability of the offered label- object pairings across several occurrences. Preference over certain referents compared to others during such associations is additionally driven by the type of the proposed referents. Referent selection and categorisation is, for example, influenced by shape, as objects with similar shapes are stored under the same label. Importantly, learning word-object associations entails further processes, such as word segmentation, object categorisation, word-order association, and recognition of morphological categories (Rubenstein & Rakic, 2013).

Where older children are concerned, the formation of links between words and referents follows more socially-driven factors, such as the referential cues provided by a speaker, based on the “principle of relevance”, whereby we tend to stimuli selectively depending on the prospective relevance and gains they offer (Henderson, Sabbagh & Woodward, 2012). Furthermore, the acquisition of novel words has been said to occur through a process termed “inferential word learning”, which takes place by making use of contextual and morphosyntactic cues (Spencer, Thompson, Petersen & Restrepo, 2018) and is highly related to the number of exposures to specific structures by a child. Moreover, the selection of a possible word-referent may also rely significantly on pragmatic cues, such as eye-gaze within an intention-reading context (Grassman, Stracke & Tomasello, 2009). On the other hand, in order to discover new meanings children are thought to make use of specific semantic constraints, one of which is assumed to prevent them from connecting a novel word with a concept that already has a name (Clark & McWhinney, 1987). This constraint, which is referred to, as “mutual exclusivity”

constraint, is further analysed later in this chapter due to its relevance to the theoretical debate on what drives bilingual semantic acquisition.

Other known constraints used by children are the so-called, whole object bias, noun bias and taxonomic bias (Halberda, 2002; Gentner, 1982; Markman, 1991). Word learning is facilitated by a semantic constraint, referred to as the noun bias, whereby certain syntactic forms (i.e., nouns) are more easily learned than words that refer to other syntactic categories, such as adjectives or verbs. Gentner (1982) provides an interesting insight of this process, pointing out that one of the most challenging tasks young children need to overcome, is not just to understand the perceptual relationships among the objects that they perceive but, also, to be able to put these relationships into words and categorise them; Gentner was a supporter of the Natural Partitions account, based on which nouns are more easily learned because they are based on object-referent connections that are much more transparent from a cognitive-perceptual point of view, compared to other predicate words, such as verbs that may share a much more ‘loose’ set of possible meanings. Crosslinguistic data of children’s initial higher frequency rates of acquisition of nouns compared to verbs has been used as evidence for this account.

(30)

12 The whole object bias is used by infants and young children to rule out irrelevant meanings by

focusing their attention on the entirety of an object that is associated with a novel referent rather than its parts or other properties. Once children have learned to associate novel words to (whole) new objects, they need to be able to use this same label with similar objects. This ability is thought to be driven by the so-called taxonomic constraint, whereby children tend to focus their attention on objects sharing the same taxonomic category rather than other possible links, such as thematic relationships that are otherwise quite commonly employed by young children, for instance during classification tasks (Markman, 1991). Despite the perceptual saliency which the thematic organisation of objects presents for young children, when they consider that they are faced with a new word or label they disregard this in favour of taxonomic organisation. Among other studies, this hypothesis was tested by presenting 4-5-year-old children with picture categorisation tasks and assigning them into two

conditions: one where a label for the pictured object was offered and one where it was not; the results confirmed the hypothesis that when no label was provided children tended to categorise pictures based on thematic relations but when they heard a new label referring to the pictured object, they tended to make taxonomic decisions (Markman, 1991).

Although using the whole object constraint is considered paramount in order to achieve object-referent associations, at the same time is thought to pose an issue to young children; apart from an object’s name, they need to learn, also, to label the object’s parts, as well as its’ other properties. This problem is thought to be bypassed through the use of the mutual exclusivity contrast, which be analysed in the next section.

b. Theoretical accounts of bilingual semantic acquisition

As with overall linguistic development, there has been a long debate regarding whether lexical development of bilingual children, specifically, is the product of two different systems or one, and when this separation takes place.

Theoretical accounts regarding the development of lexical systems include: a) the interdependence view (already presented as Unitary System Hypothesis), whereby semantic representations in each language affect each other until a specific point in development, implying that bilingual development differs from monolingual one, versus b) the autonomous approach (Separate Systems Hypothesis) which views each system as autonomous, suggesting that development would be comparable to those of monolinguals (Genesee & Paradis, 1996).

As presented below, the assumption that children use semantic constraints during the discovery of new meanings, such as the ones presented earlier (Clark, 1993) has been used in studies aiming to test the hypothesis of interdependence of the two systems (unitary hypothesis). The principle proposed by

Références

Documents relatifs

The best theory appeared to be pure phonological theory, but one parameter of this theory was not validated (salience) and the results obtained could not be fully explained by

Maillart and Parisse (2006) found out that French children with specific language impairment (SLI) presented strong difficulties in phonology when compared with normally-

Distinguish and contrast between the different methodological and theoretical models applied to the academic study of the acquisition, teaching and use of English as a second

Distingir i contrastar els diferents models teòrics i metodològics aplicats a l'estudi acadèmic de l'adquisició, l'ensenyament i l'ús de l'anglès com a segona llengua en

b Central vowels /̣‫ڴ‬/ and /є‫ڴ‬/ in open rimes As mentioned in part 3.1.2.1, Sidwell found out in Haupers’ data, that /̣‫ڴ‬/ and /є‫ڴ‬/ occur in loan words

The article represents the results of theoretical analysis of content on the subject of its personal- and didactic-definite orientation, as well as some aspects of

Dynamics of total frequencies of various groups of words in the 1800 and 2000 cores The overall frequency for the words remaining in the core during these two centu- ries

The experiments of pattern recognition which shown that the combination of second-order neurons and dynamic receptive fields allows to reduce the generalization error are described