Dynamic programming has been widely used to solve various kinds of optimization problems.In this work, we show that two crucial problems in video-based sign language and gesture recognition systems can be attacked by dynamic programming with additional multiple observations. The first problem occurs at the higher (sentence) level. Movement epenthesis [1] (me), i.e., the necessary but meaningless movement between signs, can result in difficulties in modeling and scalability as the number of signs increases. The second problem occurs at the lower (feature) level. Ambiguity of hand detection and occlusion will propagate errors to the higher level. We construct a novel framework that can handle both of these problems based on a dynamic programming approach. The me has only be modeled explicitly in the past. Our proposed method tries to handle me in a dynamic programming framework where we model the me implicitly. We call this enhanced Level Building (eLB) algorithm. This formulation also allows the incorporation of statistical grammar models such as bigrams and trigrams. Another dynamic programming process that handles the problem of selecting among multiple hand candidates is also included in the feature level. This is different from most of the previous approaches, where a single observation is used. We also propose a grouping process that can generate multiple, overlapping hand candidates. We demonstrate our ideas on three continuous American Sign Language data sets and one hand gesture data set. The ASL data sets include one with a simple background, one with a simple background but with the signer wearing short sleeved clothes, and the last with a complex and changing background. The gesture data set contains color gloved gestures with a complex background. We achieve within 5% performance loss from the automatically chosen me score compared with the manually chosen me score. At the low level, we first over segment each frame to get a list of segments. Then we use a greedy method to group the segments based on different grouping cues. We also show that the performance loss is within 5% when we compare this method with manually selected feature vectors.

PhD (Tshivenda) / Senthara ya M.E.R. Mathivha ya Nyambo dza Afrika kha Vhutsila na Mvelele / Tshivenḓa is one of African languages spoken in the Northern parts of South Africa and some parts of Zimbabwe. This proposal focuses on a study of selected phonological processes involving Tshivenḓa vowels. The Tshivenḓa phonological processes involving vowels under this study include vowel elision/deletion, vowel epenthesis, vowel coalescence, vowel nasalisation, vowel harmony, vowel raising and glide formation. In this regard, consonants are not part and parcel of this study. Furthermore, the study enriches the phonology of Tshivenḓa with more phonological processes thereby adding on the existing literature. The study also attempts to motivate and encourage Tshivenḓa speaking people and those who are interested in the language to have a clear understanding of phonological processes which occur in Tshivenḓa. The theoretical framework underpinning this study is the Optimality Theory (OT). The study will be carried out through phonological analysis, that is, field work and word list. Secondary sources will also be used for data collection.

This study is a synchronic derivational analysis of phonological phenomena in three dialects of Sudanese Arabic. Its main goal is to provide a unified prosodic account of syncope and of the phonological processes functioning as strategies of repairing unsyllabified segments in the dialects of Urban Central Sudanese Arabic, Shukriiya, and Hamar. The domains of these processes are argued to follow from the degree of restriction that dialects place on word-level and phrase-level syllabification. To this end, the study proposes an analysis of syllabification in the three dialects that identifies the degree to which word-level syllabification is exhaustive, the segments that may be marked extrasyllabic and the conditions regulating their extrasyllabic status, the phrasal level at which these segments must be syllabified, and the level at which alteration to syllable structure is disallowed. In identifying the degrees of restriction dialects place on syllabification and resyllabification, the analysis provides a principled explanation for the levels of repair of unsyllabified segments as well as the domains of syncope. The study also provides an analysis of word stress and an analysis of phonological phrase formation. By revealing and accounting for the interesting phonological patterns attested in these dialects, the study aims to contribute to the area of Arabic phonology in general and to research on the typology of Arabic dialects in particular. In addition to the analyses proposed, its substantial contribution in this regard is a significant body of original data that is being analysed for the first time. With respect to dialects of Sudanese Arabic, the study represents a new direction of enquiry, one that seeks to disentangle their respective grammars and reveal the interesting ways in which they pattern alike and diverge.

Made available in DSpace on 2015-05-14T12:43:04Z (GMT). No. of bitstreams: 1 arquivototal.pdf: 1347135 bytes, checksum: 57037afb5540568a2999801a534a4cbf (MD5) Previous issue date: 2014-02-24 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - CAPES / This study aims to describe the variable process of inserting the epenthetic vowel [ɪ] in words beginning with clusters at onset position in English by Brazilian, from Paraíba, learners of English so called foreign language. Based on research conducted regarding this phenomenon as: Escartín (2005) with Hispanic learners of English so called foreign language and similar phenomena Cardoso (2004, 2008, 2009). The objective of this research is, then, to identify the incidence frequency of the insertion of vowel epenthesis at onset position of words in English that begin with one of the following clusters /sp/, /st/, /sk/, /sl/, /sm/, and /sn/ by Brazilian learners of English so called foreign language. This work aims to answer the following guiding questions: what is the incidence frequency of the onset vowel epenthesis in sC cluster in English by Brazilian learners of English so called foreign language?; what is the role of the sonority of the second consonant of the cluster and the preceding phonological context on the occurrence of onset vowel epenthesis in sC cluster?; what is the role of the learner's proficiency, and level of output formality in the occurrence of onset vowel epenthesis in sC cluster? and what is the role of the English so called foreign language learner phonological awareness in the occurrence of onset vowel epenthesis in sC cluster? The hypothesis are that obstruent and liquid consonants can promote vowel epenthesis, whereas nasal ones do not favor it; that preceding consonantal and pause environments favor initial vowel epenthesis while preceding vowel epenthesis do not favor such occurrence; that the more proficient is the informant in the target language, the less use of initial vowel epenthesis in sC clusters; and the more formal is the data collection style less initial vowel epenthesis in sC clusters happens. The research corpus is composed of 18 informants from Paraiba, learners of English so called foreign language, distributed in basic, intermediate and advanced levels of proficiency. For speech data collection was recorded twenty-eight sentences and text in English read by the participants. The collected material was quantitatively analyzed by the computer program GOLDVARB X (Sankoff, Tagliamonte and Smith, 2005). The independent variables are learner s level of proficiency and phonological awareness in the so called foreign language, sonority of the second element of the cluster, preceding phonological context and instrument type of data collection. The statistical analysis shows, in decreasing order of significance, that the sonority of the cluster, the learner s level of proficiency, the learner s phonological awareness, and the preceding context of the cluster are the variables more relevant to the realization of onset vowel epenthesis in sC cluster. The achieved results contribute not only to understand how English so called foreign language learning occurs by Brazilian learners, but also promotes pedagogical implications for teaching English so called foreign language. / epentética em palavras iniciadas por cluster em posição inicial na língua inglesa por aprendentes brasileiros, do estado da Paraíba, de inglês como língua dita estrangeira, tendo como base pesquisas realizadas a respeito desse fenômeno como: Escartín (2005) com aprendentes hispânicos de inglês como LDE e fenômenos semelhantes Cardoso (2004, 2008, 2009). O objetivo dessa pesquisa é, então, identificar a frequência da ocorrência de inserção da epêntese vocálica ou vogal de apoio na posição inicial das palavras em língua inglesa que se iniciam com um dos seguintes clusters /sp/, /st/, /sk/, /sl/, /sm/, e /sn/ por aprendentes brasileiros de inglês como LDE. Este trabalho tem como meta responder às seguintes perguntas norteadoras: qual a frequência de ocorrência da epêntese vocálica inicial em cluster sC em L2 por aprendentes brasileiros de inglês como LDE?; qual o papel da sonoridade da segunda consoante do cluster e do contexto fonológico precedente na ocorrência de epêntese vocálica inicial em cluster sC?; qual o papel da proficiência do aprendente, e nível de formalidade da produção na ocorrência de epêntese vocálica inicial em cluster sC?; e qual o papel da consciência fonológica do aprendente de inglês como LDE na ocorrência de epêntese vocálica inicial em cluster sC? As hipóteses são que as consoantes obstruintes e líquidas possam favorecer a epêntese vocálica, enquanto que as consoantes nasais não favoreçam; o contexto fonológico precedente consonantal e de pausa favoreçam a epêntese vocálica inicial em clusters sC, ao passo que o contexto vocálico precedente desfavoreça a ocorrência de tal fenômeno; quanto mais proficiente na língua alvo for o informante, menos se utilize da epêntese vocálica inicial em cluster sC; e quanto mais formal for o tipo de coleta de dados, menos aconteça o fenômeno da epêntese vocálica em cluster sC. O corpus da pesquisa é constituído por 18 informantes paraibanos, aprendentes de inglês como LDE, pertencentes aos níveis de proficiência na LDE: básico, intermediário e avançado. Para a coleta dos dados de fala foi gravada a leitura de vinte e oito frases e um texto em inglês. O material coletado foi analisado quantitativamente através do programa computacional GOLDVARB X (SANKOFF; TAGLIAMONTE; SMITH, 2005). As variáveis independentes observadas são: nível de proficiência e consciência fonológica do aprendente, sonoridade do segundo elemento do cluster, contexto fonológico precedente e tipo de instrumento de coleta de dados. O tratamento estatístico realizado mostra, em ordem decrescente de significância, que as variáveis sonoridade do cluster, nível de proficiência, consciência fonológica do informante, contexto precedente ao cluster são as que se mostram mais relevantes à realização da epêntese vocálica inicial em cluster sC. Os resultados alcançados contribuirão não só para entender como ocorre a aprendizagem de inglês como LDE por aprendentes brasileiros, mas promove também implicações pedagógicas no ensino de inglês como LDE.

No description available.

This thesis analyzes the phonological and morphological nativisation of English loans in the Tonga language. The contact situation between English and Tonga, in Zimbabwe, facilitates transference of lexical items between the two languages. From having been one of the most widely used languages of the world, English has developed into the most influential donor of words to other languages such as Tonga. The infiltration of English words into the Tonga lexical inventory led to the adoption and subsequent nativisation of English words by the native Tonga speakers. The main deposit of English words into Tonga is the direct interaction between English and Tonga speakers. However, it is sometimes via other languages like Shona, Ndebele, Venda and Shangani. In the 21st century, English’s contribution to the vocabulary of Tonga became more widely spread, now covering a large proportion of the Tonga language’s lexical inventory. The fact that English is the medium of instruction, in Zimbabwe, language of technology, education, media, new administration, health, music, new religion and economic transactions means that it is regarded as the high variety language with coercive loaning powers. Words from English are then adopted and nativised in the Tonga language, since Tonga asserts itself an independent language that can handle loans on its own. The main focus of this study therefore, is to try and account for the phonological and morphological behavior and changes that take place in English words that enter into Tonga. Analyzing phonological processes that are employed during nativisation of loan words entails analyzing how Tonga speakers handle aspects of English language such as diphthongs, triphthongs, cluster consonants, CVC syllable structure and sounds in repairing unacceptable sequences in Tonga. The research also accounts for the handling of morphological differences between the two languages. This entails looking at how competence and ordered-rule framework are harmonized by Tonga speakers in repairing conflicting features at morphological level. Since the two languages have different morphological patterns, the research analyzes the repairing strategies to handle singular and plural noun prefixes, tenses and particles, which are morphological components of words. The researcher appreciates that the native Tonga speakers have robust intuitions on the proper way to nativise words. / African Languages / D.Litt. et Phil. (African Languages)

This thesis analyzes the phonological and morphological nativisation of English loans in the Tonga language. The contact situation between English and Tonga, in Zimbabwe, facilitates transference of lexical items between the two languages. From having been one of the most widely used languages of the world, English has developed into the most influential donor of words to other languages such as Tonga. The infiltration of English words into the Tonga lexical inventory led to the adoption and subsequent nativisation of English words by the native Tonga speakers. The main deposit of English words into Tonga is the direct interaction between English and Tonga speakers. However, it is sometimes via other languages like Shona, Ndebele, Venda and Shangani. In the 21st century, English’s contribution to the vocabulary of Tonga became more widely spread, now covering a large proportion of the Tonga language’s lexical inventory. The fact that English is the medium of instruction, in Zimbabwe, language of technology, education, media, new administration, health, music, new religion and economic transactions means that it is regarded as the high variety language with coercive loaning powers. Words from English are then adopted and nativised in the Tonga language, since Tonga asserts itself an independent language that can handle loans on its own. The main focus of this study therefore, is to try and account for the phonological and morphological behavior and changes that take place in English words that enter into Tonga. Analyzing phonological processes that are employed during nativisation of loan words entails analyzing how Tonga speakers handle aspects of English language such as diphthongs, triphthongs, cluster consonants, CVC syllable structure and sounds in repairing unacceptable sequences in Tonga. The research also accounts for the handling of morphological differences between the two languages. This entails looking at how competence and ordered-rule framework are harmonized by Tonga speakers in repairing conflicting features at morphological level. Since the two languages have different morphological patterns, the research analyzes the repairing strategies to handle singular and plural noun prefixes, tenses and particles, which are morphological components of words. The researcher appreciates that the native Tonga speakers have robust intuitions on the proper way to nativise words. / African Languages / D.Litt. et Phil. (African Languages)

Cette grammaire du tanimuka, langue tucano d’Amazonie colombienne, s’organise en quatre parties. La première partie, ‘le tanimuka : la langue et ses locuteurs’, est divisée en trois chapitres (1 à 3), qui établissent le profil sociolinguistique de la langue, et décrivent les conditions d’enquête de terrain. La deuxième partie, divisée en deux chapitres (5 et 6), présente une description de la phonologie et morphophonologie de la langue, avec une description du phénomène d’épenthèse syllabique dans le verbe. La troisième partie (chapitres 7 et 8) présente d’abord une étude générale de la grammaire au niveau morphosyntaxique et syntaxique, puis une étude thématique approfondie des systèmes de classification nominale : genre et classificateurs numéraux. La quatrième partie, divisée en trois chapitres (9 à 11), se concentre sur le domaine de l’expression spatiale, omniprésente dans la langue. Le premier chapitre présente l’inventaire de tous les éléments de l’expression spatiale de la langue. Le deuxième décrit le fonctionnement des verbes de postures et des positionnels dans les constructions locatives de base de prédicats de base simple ou complexe (séries verbales). Le troisième est dédié à l’étude de l’expression de la trajectoire. La thèse se termine par trois annexes : une présentation des stimuli visuels créés pour les enquêtes sur les verbes locatifs de posture et les positionnels, une étude sur le système de numération traditionnel du tanimuka, et la transcription d’une histoire de tradition orale ‘La tortue de terre Parekika’, glosée et traduite. / This grammar of Tanimuka, a Tucanoan language of the Colombian Amazonian area, is organized in four parts. The first part, ‘Tanimuka : the language and its speakers’, is divided into three chapters (1 to 3), which establish the sociolinguistic profile of the language and describe the fieldwork conditions. The second part, divided into two chapters (5 and 6), presents a description of the phonology and morphophonology of the language, with a study of the phenomenon of syllabic epenthesis in the verb. The third part (chapters 7 and 8) begins with a general study of the morphosyntactic and syntactic levels, followed by an in-depth study of the noun classification systems: gender and numeral classifier. The fourth part (chapters 9 to 11) focuses on the expression of space, omnipresent in the language. The first chapter presents an inventory of all the elements that express the domain of space in the language. The second provides a detailed analysis of postures and positional verbs in basic locative construction of simple or complex predicates (serial verb constructions). The final chapter is dedicated to the study of the expression of the Trajectory category. The dissertation ends with three annexes, which present, in turn, the visual stimuli created for the investigation of locative verbs of posture and position, a study of the traditional system of numeration and the transcription of a history of oral tradition: 'The tortoise Parekika', glossed and translated.