Despite the plethora of SLA research conducted on the acquisition of the definite marker in noun phrase configurations in L2 Arabic and English (e.g., Sarko, 2007; Master, 1997; Collier, 1987; Anderson, 1984; Kharma, 1981), there is as yet no definitive description of how noun phrases are acquired and why errors persist after advanced stages in L2 learning. Results, as shown by Butler (2002), are inconclusive, and the primary causes of difficulties in the acquisition of the definite marker in noun phrase configurations remain unclear. Recently, the internal syntax-semantics interface (Cuza & Frank, 2011; Montrul, 2010; Tsimpli & Sorace, 2006; Sorace, 2003, 2004) and the specificity-definiteness distinction (Ionin, 2003; Ionin et al., 2004; Ionin et al., 2008) have been considered as appropriate frameworks for exploring the acquisition of noun phrases and other structural features. The structure of noun phrase configurations in Arabic and English offers a complex interface between form and meaning for L2 learners with multiple cases of matches and mismatches between specificity and definiteness. In this three-article dissertation project, two of which were conducted in a bidirectional methodological framework with L1 Arabic-L2 English and L1 English-L2 Arabic learners, I explored the acquisition of three cases of noun phrase configurations. In the first study, I investigated the acquisition of plural noun phrase configurations that carry generic and specific readings at the initial state of L2 learning. Using three data collection instruments: written translation; error detection and correction; and forced choice elicitation, I tested the predictions made by the Interface Hypothesis (IH) and the Full Transfer (FT) Hypothesis. Results showed that L2 learners in both directions tend to transfer noun phrase configurations from L1 into L2, a result that I took to support the FT hypothesis. In addition, it took L1 English-L2 Arabic learners two years of instruction to recover from this L1 effect. The second study aimed at confirming the result of the first study, but in the acquisition of the definite marker in generic singular noun phrase configurations in the L1 English-L2 Arabic direction. The behavior of generic singular noun phrases in L2 Arabic offers a good testing ground since it has numerous similarities and differences with English. Two conditions were established: a matching condition and a mismatching condition. Both conditions were tested in the L1 English-L2 Arabic direction. Results showed a similar pattern to the one recorded in the first study. Typological proximity and distance were found to be important determiners of language acquisition of the in/definiteness configurations of singular noun phrases. In the third study, I shifted to the exploration of a more complex type of noun phrases; namely the definite Iḍāfah construction in Modern Standard Arabic (MSA) and its equivalent noun phrase configurations in English in the two directions. I started with a common difference between MSA and English. Whereas in MSA there is a canonical configuration in terms of head-complement ordering and head-complement definiteness, English is tolerant of more than one permissible configuration. I operationalized the acquisition of these noun phrases in terms of head-complement ordering and head-complement definiteness. Results showed a clear effect of L1 transfer in both directions; knowledge of L1 noun phrase configurations acts as the initial step in L2 learning. I concluded that both communities of L2 learners face problems that vary according to the L1 noun phrase configuration at hand. However, in the L1 English-L2 Arabic direction, learners reached a satisfactory level of performance in the Iḍāfah construction after two years of instruction. I approached this finding as a result of intensive Focus-on-Form Episodes (Loewn, 2005) that the Iḍāfah construction receives in MSA instruction. At the conclusion of this research project I highlighted some implications for the second language acquisition and teaching of noun phrases. The overall results were couched in a broader perspective that characterizes the initial state of L2 learning of noun phrases in + article and – article languages, the effects of typological proximity and distance, and the effects of Instructed SLA. For the pedagogical implications, I called for the integration of the semantics of the definite marker while presenting noun phrases in textbooks. I also recommended the use of explicit instruction and structured-input activities (VanPatten, 2004; Marsden & Chen, 2011) as effective pedagogical tools that foster form-meaning mapping in the acquisition of L2 Arabic and English noun phrases.

In recent years, attempts to simplify the grammatical mechanisms used in syntax have led to proposals to reduce the relationships between elements in a sentence to relations between heads and complements, doing away with free adjunction. For the analysis of modifying relations one consequence has been the rise of analyses that use the properties of selecting heads to stipulate unexpected syntactic behaviour, such as the use of light verbs to derive transparency in complex verb constructions. This thesis shows that such accounts are empirically inadequate and argues that the relationship between heads and adjuncts provides a more empirically-satisfactory model of modifying relations, such as complex verb constructions, than one restricted to the selection relation between heads and complements in the syntax. In support of the adjunct relation, I show how a modular approach to adjuncts in which the position of adjunction is licensed in the semantics and long-distance dependencies are licensed in the syntax can provide a more unified account of subextraction from two separate types of island configurations, viz. asymmetric subextraction from coordination and subextraction from participial adjuncts, either than analyses involving complementation in the syntax (Borgonovo and Neeleman, 2000; Fabregas and Jiménez-Fernández, 2016; Wiklund, 2007), or hybrid analyses mixing processing filters with syntactic licensing of long-distance dependencies (Truswell, 2009, 2011). The first part of the thesis shows that Chomsky’s (2000; 2001) phase theory gives rise to blackholes in the specifier positions of phases from which movement cannot take place. I provide a theoretical account in terms of feature-licensing, where blackholes are formed by the impossibility of licensing at least one unlicensed feature on a phase head, and show how this account derives the distinction between canonical adjuncts from which subextraction is not permitted and subextraction from single event constructions in which subextraction is permitted. The section speculatively concludes with a demonstration of how blackholes might provide a unified analysis of islandhood in general. The second part of the thesis concentrates on the empirical phenomenon of subextraction from coordination and participial adjuncts. I report the results of a series of judgement experiments run in parallel across two sets of constructions, coordination and participial adjuncts, in three languages, English, German and Norwegian. The aim was to test whether acceptability of subextraction from within coordination and participial adjuncts varied depending on the aspectual or grammatical type of matrix predicate. The results show that acceptability of subextraction does depend on the type of matrix predicate. The crucial factor is intransitivity, partially confirming the bias towards unaccusatives in subextraction from participial adjuncts observed informally in Borgonovo and Neeleman (2000); Fabregas and Jiménez-Fernández (2016); Truswell (2011) whilst providing evidence against theoretical accounts that rely primarily on unaccusativity (Borgonovo and Neeleman, 2000; Fabregas and Jiménez-Fernández, 2016), primarily on aspectual distinctions (Truswell, 2007b) or primarily on agentivity (Truswell, 2009, 2011). Interestingly, the hierarchy in acceptability between the four types of matrix predicates stays constant across all three languages, despite both pseudocoordination and subextraction from within participials being ungrammatical in German.

`Kinds of objects and their meanings´ deals with objecthood in Kiswahili. From a syntactic point of view, there is but one kind of object: the distinction between `direct´ and `indirect´ object has no syntactic properties, and one verb can have only one object. Of course, objects can have different semantic roles. This raises questions about the syntactic and semantic functions of `naked´ non-objects, and some of these are approached by inspecting fifty examples of the verb kutia `to put [sth] [into]´ from Sacleux´s dictionary. Three syntactic and semantic frames are distinguished and the respective roles of the arguments are described. Finally, there is a brief discussion about the meaning of the object as such and how it is influenced by the presence of the applicative extension.

GenDR est un réalisateur de texte symbolique qui prend en entrée un graphe, une repré- sentation sémantique, et génère les graphes sous forme d’arbres de dépendances syntaxiques lui correspondant. L’une des tâches de GenDR lui permettant d’effectuer cette transduction est la lexicalisation profonde. Il s’agit de choisir les bonnes unités lexicales exprimant les sémantèmes de la représentation sémantique d’entrée. Pour ce faire, GenDR a besoin d’un dictionnaire sémantique établissant la correspondance entre les sémantèmes et les unités lexi- cales correspondantes dans une langue donnée. L’objectif de cette étude est d’élaborer un module de lexicalisation souple construisant automatiquement un dictionnaire sémantique du français riche pour GenDR, son dictionnaire actuel étant très pauvre. Plus le dictionnaire de GenDR est riche, plus sa capacité à paraphra- ser s’élargit, ce qui lui permet de produire la base de textes variés et naturels correspondant à un même sens. Pour y parvenir, nous avons testé deux méthodes. La première méthode consistait à réorganiser les données du Réseau Lexical du Français sous la forme d’un dictionnaire sémantique, en faisant de chacun de ses noeuds une entrée du dictionnaire et des noeuds y étant reliés par un type de lien lexical que nous appelons fonctions lexicales paradigmatiques sémantiquement vides ses lexicalisations. La deuxième méthode consistait à tester la capacité d’un modèle de langue neuronal contextuel à générer des lexicalisations supplémentaires potentielles correspondant aux plus proches voisins du vecteur calculé pour chaque entrée du dictionnaire afin de l’enrichir. Le dictionnaire construit à partir du Réseau lexical du français est compatible avec GenDR et sa couverture a été considérablement élargie. L’utilité des lexicalisations supplémentaires générées par le modèle neuronal s’est avérée limitée, ce qui nous amène à conclure que le modèle testé n’est pas tout à fait apte à accomplir le genre de tâche que nous lui avons de- mandée. / GenDR is an automatic text realiser. Its input is a graph; a semantic representation, and its output is the corresponding syntactic dependencies tree graphs. One of GenDR’s tasks to operate this transduction successfully is called deep lexicalization, i.e. choosing the right lexical units to express the input semantic representation’s semantemes. To do so, GenDR needs access to a semantic dictionnary that maps the semantemes to the corresponding lexical units in a given language. This study aims to develop a flexible lexicalization module to build a rich French semantic dictionary automatically for GenDR, its current one being very poor. The more data the semantic dictionary contains, the more paraphrases GenDR is able to produce, which enables it to generate the basis for natural and diverse texts associated to a same meaning. To achieve this, we have tested two different methods. The first one involved the reorganization of the French Lexical Network in the shape of a semantic dictionary, by using each of the network’s nodes as a dictionary entry and the nodes linked to it by a special lexical relationship we call semantically empty paradigmatic lexical functions as its lexicalizations. The second method involved testing a contextual neural language model’s ability to gen- erate potential additional lexicalizations by calculating the vector of each of the dictionary entries and generating its closest neighbours in order to expand the semantic dictionary’s coverage. The dictionary we built from the data contained in the French Lexical Network is com- patible with GenDR and its coverage has been significantly broadened. Use of the additional lexicalizations produced by the language model turned out to be limited, which brings us to the conclusion that the tested model isn’t completely able to perform the task we’ve asked from it.

La Génération Automatique de Langue Naturelle vise à produire des textes dans une langue humaine à partir d'un ensemble de données non-linguistiques. Elle comprend généralement trois sous-tâches principales: (i) sélection et organisation d'un sous-ensemble des données d'entrée; ii) détermination des mots à utiliser pour verbaliser les données d'entrée; et (iii) regroupement de ces mots en un texte en langue naturelle. La dernière sous-tâche est connue comme la tâche de Réalisation de Surface (RS). Dans ma thèse, j'étudie la tâche de RS quand les données d'entrée sont extraites de Bases de Connaissances (BC). Je présente deux nouvelles approches pour la réalisation de surface à partir de bases de connaissances: une approche supervisée et une approche faiblement supervisée. Dans l'approche supervisée, je présente une méthode basée sur des corpus pour induire une grammaire à partir d'un corpus parallèle de textes et de données. Je montre que la grammaire induite est compacte et suffisamment générale pour traiter les données de test. Dans l'approche faiblement supervisée, j'explore une méthode pour la réalisation de surface à partir de données extraites d'une BC qui ne requière pas de corpus parallèle. À la place, je construis un corpus de textes liés au domaine et l'utilise pour identifier les lexicalisations possibles des symboles de la BC et leurs modes de verbalisation. J'évalue les phrases générées et analyse les questions relatives à l'apprentissage à partir de corpus non-alignés. Dans chacune de ces approches, les méthodes proposées sont génériques et peuvent être facilement adaptées pour une entrée à partir d'autres ontologies / Natural Language Generation is the task of automatically producing natural language text to describe information present in non-linguistic data. It involves three main subtasks: (i) selecting the relevant portion of input data; (ii) determining the words that will be used to verbalise the selected data; and (iii) mapping these words into natural language text. The latter task is known as Surface Realisation (SR). In my thesis, I study the SR task in the context of input data coming from Knowledge Bases (KB). I present two novel approaches to surface realisation from knowledge bases: a supervised approach and a weakly supervised approach. In the first, supervised, approach, I present a corpus-based method for inducing a Feature Based Lexicalized Tree Adjoining Grammar from a parallel corpus of text and data. I show that the induced grammar is compact and generalises well over the test data yielding results that are close to those produced by a handcrafted symbolic approach and which outperform an alternative statistical approach. In the weakly supervised approach, I explore a method for surface realisation from KB data which does not require a parallel corpus. Instead, I build a corpus from heterogeneous sources of domain-related text and use it to identify possible lexicalisations of KB symbols and their verbalisation patterns. I evaluate the output sentences and analyse the issues relevant to learning from non-parallel corpora. In both these approaches, the proposed methods are generic and can be easily adapted for input from other ontologies for which a parallel/non-parallel corpora exists

Les constructions causatives ont fait l’objet d’une attention particulière dans les études linguistiques, non seulement parce qu’elles font partie des innombrables expressions qui composent le système langagier humain, mais parce qu'elles reflètent également la mesure dans laquelle la cognition humaine perçoit et interprète la nature de la causation. Parmi les diverses causatives prédicatives, nous accordons une attention particulière aux causatives analytiques tirées des données anglaises et mandarines – deux langues qui comptent le plus grand nombre de locuteurs au monde : les constructions cause, make, have, shi (使) et rang (让). Guidé par le programme cartographique et minimaliste de la syntaxe générative, ce travail fournit, dans un premier temps, une recension des caractéristiques sémantico-syntaxiques qui permettent d’unir ou de distinguer les causatives de l’anglais et du mandarin. Ensuite, nous passons en revue les approches existantes qui discutent de la structure sous-jacente des constructions visées afin d’examiner à quel point ces approches pourraient nous guider dans la description des données interlinguistiques. Finalement, sur la base de ces approches, nous tentons de proposer une structure syntaxique plus universelle des causatives anglaises/mandarines tout en tenant compte des particularités sémantiques et des spécificités interlinguistiques. En plus d'adopter une configuration syntaxique monoclausale qui scinde la couche vP en deux, nous insistons également sur la nécessité d’introduire un trait évènementiel [contrôle] et de déterminer la valeur sémantique des verbes impliqués dans chaque structure causative anglaise/mandarine. / Causative constructions have received particular attention in linguistic studies, not only because they are part of the countless expressions that constitute the human language system, but also because they reflect the extent to which human cognition perceives and interprets the nature of causation. Among the various predicative causatives, this study focuses on analytic causatives drawn from English and Mandarin data - two languages with the largest number of speakers in the world: the cause, make, have, shi (使) and rang (让) constructions. Driven by the cartographic and minimalist program of generative syntax, this work first provides a survey of the semantic-syntactic features that unite or distinguish English and Mandarin causatives. Next, we review existing approaches that discuss the underlying structure of the target constructions to examine how these approaches might guide us in describing cross-linguistic data. Finally, based on these approaches, we attempt to propose a more universal syntactic structure of English/Mandarin causatives while considering semantic particularities and cross-linguistic specificities. In addition to adopting a monoclausal structure that splits the vP layer in two, we also insist on the need to introduce an event feature [control] and to determine the semantic value of verbs involved in each English/Mandarin causative structure.

The benefits of Model-Driven Software Development (MDSD) and Domain-Specific Languages (DSLs) wrt. efficiency and quality in software engineering increase the demand for custom languages and the need for efficient methods for language engineering. This motivated the introduction of language families that aim at further reducing the development costs and the maintenance effort for custom languages. The basic idea is to exploit the commonalities and provide means to enable systematic variation among a set of related languages. Current techniques and methodologies for language engineering are not prepared to deal with the particular challenges of language families. First, language engineering processes lack means for a systematic analysis, specification and management of variability as found in language families. Second, technical approaches for a modular specification and realisation of languages suffer from insufficient modularity properties. They lack means for information hiding, for explicit module interfaces, for loose coupling, and for flexible module integration. Our first contribution, Feature-Oriented Language Family Engineering (LFE), adapts methods from Software Product Line Engineering to the domain of language engineering. It extends Feature-Oriented Software Development to support metamodelling approaches used for language engineering and replaces state-of-the-art processes by a variability- and reuse-oriented LFE process. Feature-oriented techniques are used as means for systematic variability analysis, variability management, language variant specification, and the automatic derivation of custom language variants. Our second contribution, Integrative Role-Based Language Composition, extends existing metamodelling approaches with roles. Role models introduce enhanced modularity for object-oriented specifications like abstract syntax metamodels. We introduce a role-based language for the specification of language components, a role-based composition language, and an extensible composition system to evaluate role-based language composition programs. The composition system introduces integrative, grey-box composition techniques for language syntax and semantics that realise the statics and dynamics of role composition, respectively. To evaluate the introduced approaches and to show their applicability, we apply them in three major case studies. First, we use feature-oriented LFE to implement a language family for the ontology language OWL. Second, we employ role-based language composition to realise a component-based version of the language OCL. Third, we apply both approaches in combination for the development of SumUp, a family of languages for mathematical equations.:1. Introduction 1.1. The Omnipresence of Language Families 1.2. Challenges for Language Family Engineering 1.3. Language Family Engineering with Features and Role-Based Composition 2. Review of Current Language Engineering 2.1. Language Engineering Processes 2.1.1. Analysis Phase 2.1.2. Design Phase 2.1.3. Implementation Phase 2.1.4. Applicability in Language Family Engineering 2.1.5. Requirements for an Enhanced LFE Process 2.2. Technical Approaches in Language Engineering 2.2.1. Specification of Abstract Syntax 2.2.2. Specification of Concrete Syntax 2.2.3. Specification of Semantics 2.2.4. Requirements for an Enhanced LFE Technique 3. Feature-Oriented Language Family Engineering 3.1. Foundations of Feature-Oriented SPLE 3.1.1. Introduction to SPLE 3.1.2. Feature-Oriented Software Development 3.2. Feature-Oriented Language Family Engineering 3.2.1. Variability and Variant Specification in LFE 3.2.2. Product-Line Realisation, Mapping and Variant Derivation for LFE 3.3. Case Study: Scalability in Ontology Specification, Evaluation and Application 3.3.1. Review of Evolution, Customisation and Combination in the OWL LanguageFamily 3.3.2. Application of Feature-Oriented Language Family Engineering for OWL 3.4. Discussion 3.4.1. Contributions 3.4.2. Related Work. 3.4.3. Conclusion 4. Integrative, Role-Based Composition for Language Family Engineering 4.1. Foundations of Role-Based Modelling. 4.1.1. Information Hiding and Interface Specification in Role Models 4.1.2. Loose Coupling and Flexible Integration in Role Composition 4.2. The LanGems Language Composition System 4.2.1. The Language Component Specification Language . 4.2.2. TheLanguageCompositionLanguage 4.2.3. TechniquesofLanguageComposition 4.3. Case Study: Component-based OCL 4.3.1. Role-Based OCL Modularisation 4.3.2. Role-Based OCL Composition 4.4. Discussion 4.4.1. Contributions 4.4.2. Related Work 4.4.3. Conclusion 5. LFE with Integrative, Role-Based Syntax and Semantics Composition 5.1. Integrating Features and Roles 5.2. SumUp Case Study 5.2.1. Motivation 5.2.2. Feature-Oriented Variability and Variant Specification 5.2.3. Role-Based Component Realisation 5.2.4. Feature-Oriented Variability and Variant Evolution 5.2.5. Model-driven Concrete Syntax Realisation 5.2.6. Model-driven Semantics Realisation 5.2.7. Role-Based Composition and Feature Mapping 5.2.8. Language Variant Derivation 5.3. Conclusion 6. Conclusion 6.1. Contributions 6.2. Outlook 6.2.1. Co-Evolution in Language Families 6.2.2. Role-Based Tool Integration. 6.2.3. Automatic Modularisation of Existing Language Families 6.2.4. Language Component Library Appendix A Appendix B Bibliography