This thesis investigates the developpment of gender agreement in determiners and adjectives in the spontaneous speech of L2 French by five groups of Swedish learners: beginners at the university, secondary school students, university students, teacher candidates and PhD students. Different types of determiners are examined, such as definite and the indefinite articles. Adjectival agreement is studied in different positions in relation to the noun, such as the attributive anteposition, the attributive postposition and the predicative position. The aim is to identify the developmental sequence of gender agreement through a longitudinal study of learners at different levels of acquisition. The analysis is based on spoken language, i.e. 81 interviews belonging to the InterFra-corpus, Stockholm University. Our data also includes 8 oral productions from a control group of native speakers. The study is in three parts: one for the agreement between determiners and nouns, another for the agreement between adjectives and nouns and, finally, a study considering agreement between all three items within the noun phrase, i.e. determiner, noun and adjective (Det-N-Adj). A sequence of acquisition for gender agreement on determiners and adjectives is proposed based on the productions of four learner groups and compared to the six developmental stages suggested by Bartning and Schlyter (2004). Results have showed that there is an acquisition order of gender agreement in different parts of the nominal phrase, according to the type of determiner and the positions of the adjective. A qualitative analysis has shown a random use of gender agreement on determiners and some nouns are used with both genders on the determiner. Also, the type-token ratio is very low at the beginning of the acquisition, which partly explains the high accuracy rate (100 %). The study considering agreement between all three constituents within the noun phrase revealed that advanced learners have higher accuracy rate for gender agreement on adjectives within the noun phrase with the presence of a determiner that marks gender distinction (i.e. a non-elided, singular determiner). Results also showed that the feminine form of the adjectives remains difficult at higher acquisitional levels.

`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.

This thesis investigates the recognition rate improvement strategies for a Japanese speech recognition system. Both training data development and consonant correction scheme are studied. For training data development, a database of 995 two-syllable Japanese words is established by phonetic balanced sieving. Furthermore, feature models for the 188 common Japanese mono-syllables are derived through mixed position training scheme to increase recognition rate. For consonant correction, a sub-syllable model is developed to enhance the consonant recognition accuracy, and hence further improve the overall correct rate for the whole Japanese phrases. Experimental results indicate that the average correct rate for Japanese phrase recognition system with 34 thousand phrases can be improved from 86.91% to 92.38%.

A Mandarin phrase recognition system based on MFCC, LPC scaled excitation, vowel model, hidden Markov model (HMM) and Viterbi algorithm is proposed in this thesis. HMM, which is broadly used in speech recognition at present, is adopted in the main structure of recognition. In order to speed up the recognition time, we take advantage of stability of vowels in Mandarin and incorporate with vowel class recognition in our system. For the speaker-dependent case, a single Mandarin phrase recognition can be accomplished within 1 seconds on average in the laboratory environment.

In this work, the primary goal will be to construct the most descriptively and explanatorily adequate analysis possible to account for the complementary distribution of the Spanish copula verbs ser and estar. Over the past several decades, numerous theoretical accounts have been put forth in an attempt to accomplish this goal. Though such accounts accurately predict most types of stative sentences with the two copulas, they often fall short of predicting a significant number of them that are used in everyday speech. The first chapters of this dissertation will be devoted to reviewing a number of existing approaches that have been taken to account for the uses of ser and estar by testing their theoretical viability and descriptive adequacy. Among these are traditional conventions such as the inherent qualities vs. current condition distinction and the analysis of estar as an indicator of change. Those of a more recent theoretical framework, which will receive the most attention, include the application of Kratzer's (1995) individual-level vs. stage-level distinction to stative predicates and Maienborn's (2005) discourse-based interpretation of Spanish copulative predication. Schmitt's (2005) compositionally-based analysis of Portuguese ser and estar, which treats only estar as an aspectual copula, will be of special interest. After testing each of these analyses, it will be shown that the least costly and most accurate course to take for analyzing ser and estar is to treat both verbs as aspectual morphemes along the lines of Luján (1981). As aspectual copulas, ser and estar denote the aspectual distinction [±Perfective]. In my proposed analysis, I will argue that aspect applies to both events and states, but does so internally and externally respectively. By adapting Verkuyl's (2004) feature algebra to states, I will posit that aspect for stative predication is compositionally calculated, and the individual aspectual values for ser and estar remain constant in co-composition. In light of its descriptive adequacy for Spanish stative sentences and universality in natural language, it will also be shown that the [±Perfective] aspectual distinction is very strong in terms of explanatory adequacy as well. / text

Code-switching (CS) between French and Arabic is common across North Africa and in parts of the Middle East. Many researchers have examined this phenomenon in Tunisia (Belazi 1991, Lawson & Sachdev 2000, Belazi et. al 1994) and Morocco (Abbassi 1977, Bentahila 1983, Bentahila & Davies 1983, Lahlou 1991, Redouane 2005.) Corpus and elicited data from these two countries has helped form the basis of proposed universal constraints on code-switching, specifically the Functional Head Constraint (FHC) (Belazi et al 1994) and the Complement Adjunct Distinction (CAD) (Mahootian and Santorini 1996). However, CS between French and Moroccan and Tunsian dialects has not been directly investigated within a single study. This study is a step in filling that gap. Using a web-based survey, the present study examines native dialect speakers’ ratings of authenticity of sentences that contain both French and Arabic with a switch occurring in the Determiner Phrase (DP). The syntactic structure of the DP in the dialects examined is the same, (DP = D (D) N (A)). This is similar to the DP in French (DP = D (A) N (A)) with a few key differences that make it possible to test the FHC and CAD within the DP alone. An example of one of the eight possible switch types, between an Arabic Demonstrative Determiner and a French Definite Determiner, is seen here between Moroccan Arabic and French: Men dima had l’homme n’aime pas les chiens. (Since always this the man doesn’t like dogs.) A mixed-model ANOVA performed on the participants’ ratings reveals main effects for dialect, sex and switch type. Significant interactions also exist, including an interaction between switch type, sex and dialect. While further research is needed, the results indicate that syntactic constraints may not be the only way to understand the practice of CS. Instead, a typological approach, as suggested by Muysken (2000), may lead to a more complete understanding of why and how communities use multiple languages. / text

Bitext alignment is the task of aligning a text in a source language and its translation in the target language. Aligning amounts to finding the translational correspondences between textual units at different levels of granularity. Many practical natural language processing applications rely on bitext alignments to access the rich linguistic knowledge present in a bitext. While the most predominant application for bitexts is statistical machine translation, they are also used in multilingual (and monolingual) lexicography, word sense disambiguation, terminology extraction, computer-aided language learning andtranslation studies, to name a few.Bitext alignment is an arduous task because meaning is not expressed seemingly across languages. It varies along linguistic properties and cultural backgrounds of different languages, and also depends on the translation strategy that have been used to produce the bitext.Current practices in bitext alignment model the alignment as a hidden variable in the translation process. In order to reduce the complexity of the task, such approaches suppose that a word in the source sentence is aligned to one word at most in the target sentence.However, this over-simplistic assumption results in asymmetric, one-to-many alignments, whereas alignments are typically symmetric and many-to-many.To achieve symmetry, two one-to-many alignments in opposite translation directions are built and combined using a heuristic.In order to use these word alignments in phrase-based translation systems which use phrases instead of words, a heuristic is used to extract phrase pairs that are consistent with the word alignment.In this dissertation we address both the problems of word alignment and phrase pairs extraction.We improve the state of the art in several ways using discriminative learning techniques.We present a maximum entropy (MaxEnt) framework for word alignment.In this framework, links are predicted independently from one another using a MaxEnt classifier.The interaction between alignment decisions is approximated using stackingtechniques, which allows us to account for a part of the structural dependencies without increasing the complexity. This formulation can be seen as an alignment combination method,in which the union of several input alignments is used to guide the output alignment. Additionally, input alignments are used to compute a rich set of feature functions.Our MaxEnt aligner obtains state of the art results in terms of alignment quality as measured by thealignment error rate, and translation quality as measured by BLEU on large-scale Arabic-English NIST'09 systems.We also present a translation quality informed procedure for both extraction and evaluation of phrase pairs. We reformulate the problem in the supervised framework in which we decide for each phrase pair whether we keep it or not in the translation model. This offers a principled way to combine several features to make the procedure more robust to alignment difficulties. We use a simple and effective method, based on oracle decoding,to annotate phrase pairs that are useful for translation. Using machine learning techniques based on positive examples only,these annotations can be used to learn phrase alignment decisions. Using this approach we obtain improvements in BLEU scores for recall-oriented translation models, which are suitable for small training corpora.

This thesis investigates the second language acquisition of English psych predicates by Chinese-speaking and French-speaking adult learners of English within the Government and Binding Theory. Two major parts comprise the whole work: a study of psych predicates across Chinese, English and French, including verbs like blame and annoy, adjectives such as annoying and annoyed, and nominals like annoyance; and an experiment on Chinese and French learners' knowledge of English psych predicates. / An account of psych predicates is proposed, under which Experiencer Object (EO) verbs are the causatives of Experiencer Subject (ES) verbs, derived by zero affixation. Different D-structures are suggested for the two classes of verbs, solving the linking problem of psych predicates. The binding problem with EO verbs and corresponding -ing adjectives is resolved by the assumption of anaphoric pro, which enables the anaphor to be bound backwards by the antecedent through the extension of chain-binding theory. The Target/Subject Matter (T/SM) restriction is ruled out by a generalization established on the interaction of the zero CAUS and selectional restrictions. / Given the linguistic analysis that EO verbs are made up of a zero CAUS and a root, and the fact that psych adjectives and psych nominals are derived from these verbs, the central hypothesis for the L2 acquisition of English psych predicates hinges on this zero CAUS. It is predicted that if L2 learners of English have difficulty figuring out the causative nature of EO verbs and -ing adjectives, they should have difficulty recognizing the correct argument structure, the ungrammaticality of T/SM violations and the grammaticality of backwards binding with these predicates. A picture identification task, a multiple choice task and a grammaticality judgment and correction task are designed to test L2 learners' knowledge of these properties. The results obtained through the experiment are discussed with respect to the issues in second language acquisition.

No description available.

Les sens complémentaires en arabe sont multiples et variés. Outre les aspects syntaxiques, nous relevons des aspects sémantiques propres à chaque complément. L'introduction de l'un de ces sens dans la phrase l'enrichit et entraîne des précisions qui se rattachent au verbe. Les grammairiens classiques parlent des (mafûîät) ou encore des (manûübät) noms au cas accusatif, ils s'intéressent ainsi au traits formels des compléments. Les rhétoriciens soulignent les traits sémantiques. Ils insistent sur le sens de la restriction déterminative qu'exerce le complément sur le verbe. Une approche critique démontre que les compléments en arabe jouent un rôle important dans l'éclaircissement du sens de l'action. C'est donc le verbe qui sélectionne le sens complémentaire convenable. En arabe moderne (dans les écritures contemporaines) l'usage des compléments ne s'écarte pas loin de l'usage classique. Les auteurs sont soumis aux règles de la théorie grammaticale des premiers siècles. Toutefois, il existe certains aspects d'évolution qui reviennent, soit au choix libre de l'écrivain, soit à l'évolution interne de la langue elle-même.