This thesis investigates the syntactic features, functions, and diachrony of a complex predicate called ‘the inceptive construction’ which is based on a grammaticalised use of the converbs ‘get up’, ‘pick up’, ‘grasp’, and ‘take’. The languages under investigation are Amharic, Argobba, Harari, Zay, and Selt’i. The data collection that was analized consists of elicitations, audio recordings, and written texts. The analysis shows that the converbs identify the initial phase of the event encoded by a following verb. The converbs are further associated with nuances like volition, surprise, and emphasis. The rise of such interpretations as surprise and emphasis appears to depend mainly on context, while volition is inherent to the construction. The languages generally do not show much variation. However, there is a notable difference in some co-occurrence restrictions. Moreover, there is a difference in the presence/absence of certain converbs mainly in Harari and Zay, which is clearly a matter of preference between individual consultants. Regarding the origin of the inceptive construction, collocation, frequency, and speakers’ conception of the action of the converbs are possible factors that lead the verbs to grammaticalize into markers of the inception phase. Only some traces of the construction are found in an old Amharic text from the 15th century. The converb is the main verb form used in the inceptive construction, although other verb forms are allowed which may take a coordinating conjunction (in the cases of Amharic and Argobba) and an iterative/simultaneity marker (‘while’). The Amharic conjunction =nna ‘and’ links the light verb with the reference verb in the inceptive construction, but is also used in causal(purposive) and conditional coordination. The criteria of tense iconicity and variable positions indicate that =nna is a coordinating conjunction in the former, but a subordinator in the latter. Lastly, the converb in Amharic is shown to become insubordinated, i.e. the main verb or auxiliary it depends on gets ellipsed over time and it comes to function as a main verb. An insubordinated converb is used in the expression of surprise/exclamation, interrogation, rhetorical questioning, wishing, and the resultative/perfective. The point is it is still possible to use the notion of ‘converb’ in the inceptive construction as this is a separate historical process.

This is a corpus-based study about the light verb predicates (LVP) in Gawarbati, an Indo-Aryan language indigenous to the Hindu Kush region. This thesis aims to identify and analyse the frequency of light verbs, examine the borrowed elements within these constructions, and explore the semantic fields expressed through LVP. The data are speech events, gathered for the ongoing project “Gawarbati: Documenting a vulnerable linguistic community in the Hindu Kush”(2020-01500). The speech events were extracted from a corpus using Python and analysed manually. The results show that LVP represent 5.86% of all predicates in the data, with ‘to do’, ‘to give’, and ‘to say’ being the primary verbs, while a few others occur infrequently. Borrowed non-verbal elements in LVP, originating from Arabic, English, Persian, Pashto, and Urdu, constitute 68.92% of non-verbal elements within LVP. The main semanticfields expressed with LVP are basic actions and technology, speech and language, social andpolitical relations, cognition, and agriculture and vegetation. / Detta är en korpusbaserad studie om lättverbspredikat (LVP) i gawarbati, ett indoariskt språk som talas i Hindukush-området. Uppsatsens syfte är att identifiera lättverb i LVP i gawarbatioch analysera deras frekvens, undersöka låneord i LVP, samt utforska de semantiska fält som uttrycks genom sådana konstruktioner. Datan består av talhändelser som samlades ininför projektet “Gawarbati: Documenting a vulnerable linguistic community in the HinduKush”(2020-01500). Python användes för att exportera data från korpusen och vidare analys har gjorts manuellt. Resultatet visade att LVP representerar 5,86% av alla predikat i datan, med ‘att göra’, ‘att ge’ och ‘att säga’ som de primära lättverben och ett fåtal till som förekommer sällan. Inlånade icke-verbala element inom LVP härstammande från arabiska, engelska, persiska, pashto och urdu, och utgjorde 68,92% av alla hittade icke-verbala element inom LVP. De huvudsakliga semantiska fälten som uttrycktes med LVP var basala handlingar och teknologi, tal och språk, sociala och politiska relationer, kognition och jordbruk och vegetation. / Gawarbati: Documenting a vulnerable linguistic community in the Hindu Kush (VR 2020-01500)

Cette thèse vise à traiter les verbes composés en japonais et les verbes préverbés en polonais comme faisant partie des constructions verbales à prédicat complexe. D’abord, la notion de la formation des mots et, par conséquent, la notion de « mot », sont examinées par les analyses entre les mots (ortho)graphiques et prosodiques ; le problème d’espace entre les mots et différents systèmes d’accentuation sont pris en compte. D’une part, l’analyse des verbes composés japonais remonte jusqu’aux deux premières grammaires au XVIIe siècle, celle de Rodriguez et de Collado, dont la première propose la dichotomie entre les verbes composés à « mode de l’action » et à « particule ». Ensuite suivent les analyses des verbes composés contemporains, avec un accent particulier sur un certain nombre de couples de verbes transitifs et intransitifs en second élément du composé. D’autre part, la typologie des constructions préverbales de la langue polonaise est examinée. Enfin, à travers la comparaison multilingue du Petit prince de Saint-Exupéry entre l’original et deux traductions japonaises et deux polonaises, une typologie des constructions verbales à prédicat complexe est proposée, ce qui fait écho au choix de notre langue de rédaction, le français servant de pivot afin d’examiner diverses constructions verbales de nos langues de comparaison, le japonais et le polonais : composition verbale et construction converbale pour le premier et préverbation et construction infinitivale à semi-auxiliaire pour le second. / This thesis seeks to analyse Japanese compound verbs and Polish prefixed verbs as a part of complex predicate verbal constructions. First, the notion of word-formation and consequently that of word are examined by analyses between (ortho)graphic and prosodical words, which include problems of word space and several accentual systems. Next, this analysis of compound verbs goes back to 17th century with the first two Japanese grammars by Rodriguez and Collado, respectively, where the former proposes a dichotomy between manner of action and particle verbal compounds. Then follows an analysis of contemporary Japanese compound verbs, with particular emphasis on some pairs of transitive and intransitive verbs in the second element of compound. Contrastingly, a typology of verbal prefix constructions in Polish is considered and reviewed. Finally, through a multilingual comparison of The Little Prince, by Saint-Exupéry, using two translations in each of Japanese and Polish, a typology of complex predicate verbal constructions is put forward which corresponds with the language, French, in which this thesis is written and which serves as a pivot for the examination of various verbal constructions of the languages for comparison, Japanese and Polish: verbal compounding and converbal construction for the former, verbal prefixation and infinitival auxiliary construction for the latter.

Dans cette thèse, nous présentons une analyse syntaxique du vietnamien de Hô-Chi-Minh-Ville, dans la lignée de la syntaxe fonctionnelle mise en place par André Martinet et développée par la suite par ses continuateurs. Dans un premier temps, nous nous sommes intéressée au degré d’opposition verbo-nominal en vietnamien : nous proposons tout d’abord une synthèse des travaux menés à ce sujet, avant d’exposer notre analyse en classes, en nous appuyant sur des critères strictement syntaxiques. Une étude syntaxique détaillée des modalités (déterminants non déterminables) – les modalités nominalisantes, les modalités de degré, les modalités prédicatives et les modalités injonctives – a été nécessaire. Ces unités permettent de distinguer des comportements différents, et l’identification des grandes classes lexicales du vietnamien : nominaux, adjectifs, modaux et verbaux. Il apparait que ces classes d’unités s’organisent en un continuum, avec à un pôle les nominaux, à l’autre les verbaux. Dans un second temps, l’analyse porte sur les différents types de structures prédicatives du vietnamien. La variété des constructions – nucléaires et connectives – est au centre de l’étude, ainsi que l’identification des différentes fonctions. Enfin, nous exposons une analyse des “séries verbales”, phénomène très courant en vietnamien, mais décrivant en fait des réalités très diverses. Tout au long de l’analyse, nous accordons une importance particulière à la variation et aux zones de changements (réanalyses, lexicalisations, grammaticalisations), faisant état de la dynamique des différentes structures et de la langue elle-même. / We present here a syntaxic analysis of the Vietnamese language spoken in Ho Chi Minh City, in line with the functional syntax introduced by André Martinet and developed later by his followers. In a first part, we examined the extent of noun/verb opposition in Vietnamese. After a synthesis of work on this subject, we present our own analysis of the lexical classes, relying on strictly syntactic criteria. A detailed study of the syntactic modalités (undeterminable determinants) - nominalizing undeterminable determinants, undeterminable determinants of degree, predicative undeterminable determinants and injunctive undeterminable determinants - was necessary. These units are used to distinguish different behaviors, and leads to the identification of major lexical classes in Vietnamese: nominals, adjectives, modals and ‘verbals’. It appears that these classes of units are organized along a continuum. In a second part, the analysis focuses on the different types of predicative structures in Vietnamese. The variety of structures - nuclear and connective - is central to the study as well as the identification of the different functions. Finally, we present an analysis of "serial verb constructions" very common in Vietnamese, but actually reflecting very different realities. Throughout this study, we pay particular attention to the variation and change (reanalysis, lexicalizations, grammaticalizations), referring to the dynamics of the various structures and of the Vietnamese language itself.