This thesis aims to build a system to tackle the task of diagnosing the grammatical errors in sentences written by learners of Chinese as a foreign language with the help of the CRF model (Conditional Random Field). The goal of this task is threefold:  1) identify if the sentence is correct or not, 2) identify the specific error types in the sentence, 3) find out the location of the identified errors. In this thesis, the task of Chinese grammatical error diagnosis is approached as a sequence tagging problem. The data and evaluation tool come from the previous shared tasks on Chinese Grammatical Error Diagnosis in 2016 and 2017. First, we use characters and POS tags as features to train the model and build the baseline system. We then notice that there are overlapping errors in the data. To solve this problem, we adopt three approaches: filtering out the problematic data, assigning encoding to characters with more than one label and building separate classifiers for each error type. We continue to increase the amount of training data and include syntactic features. The results show that both filtering out the problematic data and including syntactic features have a positive impact on the results. In addition, difference between domains of training data and test data can hurt performance to a large extent.

abstract: Purpose: The present study examined grammatical gender use in child Spanish heritage speakers (HSs) in order to determine whether the differences observed in their grammar, when compared to Spanish monolinguals, stem from an incompletely acquired grammar, in which development stops, or from a restructuring process, in which features from the dominant and the weaker language converge to form a new grammatical system. In addition, this study evaluated whether the differences usually found in comprehension are also present in production. Finally, this study evaluates if HSs differences are the result of the input available to them. Method: One-hundred and four typically developing children, 48 HSs and 58 monolingual, were selected based on two age groups (Preschool vs. 3rd Grade). Two comprehension and three production experimental tasks were designed for the three different grammatical structures where Spanish expresses gender (determiners, adjectives, and clitic pronouns). Linear mixed-models were used to examine main effects between groups and grammatical structures. Results: Results from this study showed that HSs scored significantly lower than monolingual speakers in all tasks and structures; however, 3rd-Grade HSs had higher accuracy than PK-HSs. Error patterns were similar between monolinguals and HSs. Moreover, the commonly reported overgeneralization of the masculine form seems to decrease as HSs get older. Conclusion: These results suggest that HSs’ do not face a case of Incomplete Acquisition or Restructured Grammatical gender system, but instead follow a protracted language development in which grammatical skills continue to develop after preschool years and follow the same developmental patterns as monolingual children / Dissertation/Thesis / Doctoral Dissertation Speech and Hearing Science 2018

Recent research has shown that language is sensitive to probabilities and a whole host of multivariate conditioning factors. However, most of the research in this arena centres on the grammar of English, and, as yet, there is no statistical modelling on the grammar of Latin, studies of which have to date been largely philological. The rise in advanced statistical methodologies allows us to capture the underlying structure of the rich datasets which this corpus only language can potentially offer. This thesis intends to remedy this deficit by applying probabilistic and multivariate models to a specific case study, namely the alternation of word order in Latin participle auxiliary clusters (pacs), which alternate between participle-auxiliary order, as in mortuus est ‘dead is’ and est mortuus ‘is dead’. The broad research questions to be explored in this thesis are the following: (i) To what extent are probabilistic models useful and reflective of Latin syntax variation phenomena?, (ii) What are the most useful statistical models to use?, (iii) What types of linguistic variables influence variation, (iv) What theoretical implications and explanations do the statistical models suggest?Against this backdrop, a dataset of 2409 pac observations are extracted from Late Re- publican texts of the first century bc. The dataset is annotated for an “information space” of thirty-three predictor variables from various levels of linguistics: text and lemma-based variability, prosody and phonology, grammar, semantics and pragmatics, and usage-based features such as frequency. The study exploits such statistical tools as generalized linear models and multilevel generalized linear models for the regression modelling of the binary categorical outcome. However, because of the potential collinearity, and the many predictor terms, amongst other issues, the use of these models to assess the joint effect of all predictors is particularly problematic. As such, the new statistical toolkit of random forests is utilized for evaluating the relative contribution of each predictor. Overall, it is found that Latin is indeed probabilistic in its grammar, and the condition- ing factors that govern it are spread widely throughout the language space. It is also noted that probabilistic models, such as the ones used in this study, have practical applications in traditional areas of philology, including textual criticism and literary stylistics.

Le point de départ de cette recherche est le constat d’une rupture dans le fonctionnement sémantico-référentiel du genre grammatical, qui oppose d’un côté noms d’inanimés et d’animés non-anthropomorphisés, et de l’autre noms d’animés humains ou anthropomorphisés. Ce constat amène inévitablement la question, souvent traitée, du type de répartition des substantifs (arbitraire ou motivé) que permet le genre grammatical. Le fait de centrer le propos sur les noms d’humains, et plus précisément, sur la classe des « dénominations de la personne », permet de sortir de cette opposition pour analyser plus précisément les difficultés posées par la catégorie grammaticale du genre dans son lien avec la bipartition sexuée des êtres humains, généralement et traditionnellement pensée comme première. Ce travail de thèse, appuyé sur les théories de la dénomination et affiné par les outils du matérialisme et des réflexions queer sur le langage, est centré sur une proposition d’analyse stéréotypique du sens du genre grammatical. Celle-ci permet à la fois de ne pas penser l’idée d’une hiérarchie entre les genres grammaticaux (« le masculin l’emporte... ») comme structurelle et interne au système linguistique français, et de comprendre certains phénomènes en apparence contradictoires et généralement rejetés comme idéologiques et/ou politiques, donc non-linguistiques. Cette hypothèse émerge d’une réflexion sur le concept de catégorisation et sur les difficultés phénoménologiques et linguistiques qui lui sont liées. La proposition avancée est de plus orientée vers le développement d’un modèle lexicographique : le travail engagé dans cette thèse de doctorat vise donc une applicabilité potentielle. / This research was initiated with the idea of a semantic and referential splitting of grammatical gender within the French language between nouns denoting inanimates or non-anthropomorphic animates, and nouns denoting human or anthropomorphic animates. This splitting inevitably leads to the traditional question of the arbitrary or motivated nature of grammatical gender. The fact that this study focuses only on nouns denoting human animates, and more specifically on person denominations, enables to surpass this question and analyze more carefully the difficulties that arise from the idea of a link between grammatical gender and sexual bipartition. My work, nourished both by denomination theories and material and queer theories on language, is thus centered on proposing a stereotype-based semantic analysis of grammatical gender. This analysis opposes the idea of a structural hierarchy between masculine and feminine grammatical genders, and enables to understand some of the phenomena that are usually not considered as linguistic, but rejected as ideological or political. This hypothesis is thus born of a discussion of categorization theories, and of the phenomenological and linguistic difficulties that they present. Finally, one of the goals of this work is to be appliable : I will thus propose a lexicographic model of the stereotype-based hypothesis.

Animals such as bees, ants, birds, fish, and others are able to efficiently perform complex coordinated tasks like foraging, nest-selection, flocking and escaping predators without centralized control or coordination. These complex collective behaviors are the result of emergence. Conventionally, mimicking these collective behaviors with robots requires researchers to study actual behaviors, derive mathematical models, and implement these models as algorithms. Since the conventional approach is very time consuming and cumbersome, this thesis uses an emergence-based method for the efficient evolution of collective behaviors. Our method, Grammatical Evolution algorithm for Evolution of Swarm bEhaviors (GEESE), is based on Grammatical Evolution (GE) and extends the literature on using genetic methods to generate collective behaviors for robot swarms. GEESE uses GE to evolve a primitive set of human-provided rules, represented in a BNF grammar, into productive individual behaviors represented by Behavior Tree (BT). We show that GEESE is generic enough, given an initial grammar, that it can be applied to evolve collective behaviors for multiple problems with just a minor change in objective function. Our method is validated as follows: First, GEESE is compared with state-of-the-art genetic algorithms on the canonical Santa Fe Trail problem. Results show that GEESE outperforms the state-of-the-art by a)~providing better solutions given sufficient population size while b)~utilizing fewer evolutionary steps. Second, GEESE is used to evolve collective swarm behavior for a foraging task. Results show that the evolved foraging behavior using GEESE outperformed both hand-coded solutions as well as solutions generated by conventional Grammatical Evolution. Third, the behaviors evolved for single-source foraging task were able to perform well in a multiple-source foraging task, indicating a type of robustness. Finally, with a minor change to the objective function, the same BNF grammar used for foraging can be shown to evolve solutions to the nest-maintenance and the cooperative transport tasks.

京都大学 / 0048 / 新制・課程博士 / 博士(人間・環境学) / 甲第20639号 / 人博第828号 / 新制||人||198(附属図書館) / 29||人博||828(吉田南総合図書館) / 京都大学大学院人間・環境学研究科共生人間学専攻 / (主査)教授 谷口 一美, 教授 齋藤 治之, 教授 壇辻 正剛, 教授 山梨 正明 / 学位規則第4条第1項該当 / Doctor of Human and Environmental Studies / Kyoto University / DGAM

The process of acquiring language requires children to learn grammatical categories and apply these categories to new words. Researchers have proposed various explanations of this process in the form of algorithms and computational modeling. Recently, adaptation and selection models have been tested and applied as a possible explanation to the process of acquiring grammatical categories. These studies have proven promising, however, the external validity of this approach has not been examined by grammatically coding samples outside the training corpus. The current thesis applies an adaptation and selection model, which pauses the evolution of dictionaries after every thousand cycles to allow the tagging of 30 outside samples, which are then checked for tagging accuracy. The accuracy across the five training corpora by the six thousandth cycle averaged 76.75%. Additional research is needed to explore the effects of altering the parameters in the model.

The present study measured the accuracy of automated grammatical tagging software as compared to manual tagging in Spanish-speaking children's personal and fictional event narrative language samples. Studies have identified articles, clitic (contracted with a verb) pronouns, and verbs as clinical markers for language impairment in Spanish-speaking children. Automated grammatical tagging software may aid in the rapid identification of these grammatical markers. Grammatical morphemes of 30 first and fourth grade children's personal and fictional event narrative samples were tagged and compared with their respective manually tagged samples. The accuracy of word-level coding averaged 91%, and similar accuracy was found for clinically significant tags. Automated grammatical analysis has the potential to accurately identify clinically relevant grammatical forms in samples from children who speak Spanish.

The purpose of this research was to study grammatical categories in the Tagalog mental lexicon using lexical decision tasks. Some linguists question whether words in Tagalog can be classified as nouns and verbs (Foley, 1998; Kaufman, 2011) because most root words can be inflected for any grammatical function and because verbs cannot be used in their uninflected form. Previous studies with English and German (Kauschke and Stenneken 2008) have shown that participants respond differently to nouns and verbs in lexical decision tasks. These studies have also shown that participants respond differently to transitive and intransitive verbs in lexical decision tasks. It was assumed that if nouns and verbs exist in Tagalog, response times to Tagalog lexical decision tasks will show similar patterns to those performed in English and German. Two experiments were performed to examine whether words are classified as nouns and verbs in the Tagalog mental lexicon and whether other factors affected that classification. For the experiments, native speakers of Tagalog participated in lexical decision tasks and response times were measured. The first experiment tested the classification of root nouns and verbs. Contrary to findings in other languages, there was no significant difference between response times to nouns and verbs. However, there were differences in response times to nouns from different semantic categories and to verbs with different morphosyntactic structures. The second experiment examined the classification of inflected nouns and verbs. Again, the results showed no difference between response times to nouns and verbs. There was also no difference between transitive and intransitive verbs. However, there was a slight difference between verbs of different voice inflections. The results of the experiments suggest the while the grammatical classes of nouns and verbs may not be the most important features of words in the Tagalog mental lexicon, they may still play a role since different features, semantics or morphosyntactics, did affect the responses to words from the different categories.

By the later preschool years, most children have a knowledge of the grammatical categories of their native language and are capable of expanding this knowledge to novel words. To model this accomplishment, researchers have created a variety of explicit, testable models or algorithms. These have had partial but promising success in extracting grammatical word categories from transcriptions of caregiver input to young children. Additional insight into children's learning of the grammatical categories of words might be gained from evolutionary computing algorithms, which apply principles of evolutionary biology such as variation, adaptive change, self-regulation, and inheritance to computational models. The current thesis applied such a model to the language addressed to five children, whose ages ranged from 1;1 to 5;1 (years;months). The model evolved dictionaries linking words to their grammatical tags and was run for 4000 cycles; four different rates of mutation of offspring dictionaries were assessed. The accuracy for coding the words in the corpora of language addressed to the children averaged 92.74%. Directions for further development and evaluation of the model are proposed.