In the world of human language technology, resource–scarce languages (RSLs) suffer from the problem of little available electronic data and linguistic expertise. The Lwazi project in South Africa is a large–scale endeavour to collect and apply such resources for all eleven of the official South African languages. One of the deliverables of the project is more natural text–to–speech (TTS) voices. Naturalness is primarily determined by prosody and it is shown that many aspects of prosodic modelling is, in turn, dependent on part–of–speech (POS) information. Solving the POS problem is, therefore, a prudent first step towards meeting the goal of natural TTS voices. In a resource–scarce environment, obtaining and applying the POS information are not trivial. Firstly, an automatic tagger is required to tag the text to be synthesised with POS categories, but state–of–the–art POS taggers are data–driven and thus require large amounts of labelled training data. Secondly, the subsequent processes in TTS that are used to apply the POS information towards prosodic modelling are resource–intensive themselves: some require non–trivial linguistic knowledge; others require labelled data as well. The first problem asks the question of which available POS tagging algorithm will be the most accurate on little training data. This research sets out to answer the question by reviewing the most popular supervised data–driven algorithms. Since literature to date consists mostly of isolated papers discussing one algorithm, the aim of the review is to consolidate the research into a single point of reference. A subsequent experimental investigation compares the tagging algorithms on small training data sets of English and Afrikaans, and it is shown that the hidden Markov model (HMM) tagger outperforms the rest when using both a comprehensive and a reduced POS tagset. Regarding the second problem, the question arises whether it is perhaps possible to circumvent the traditional approaches to prosodic modelling by learning the latter directly from the speech data using POS information. In other words, does the addition of POS features to the HTS context labels improve the naturalness of a TTS voice? Towards answering this question, HTS voices are trained from English and Afrikaans prosodically rich speech. The voices are compared with and without POS features incorporated into the HTS context labels, analytically and perceptually. For the analytical experiments, measures of prosody to quantify the comparisons are explored. It is then also noted whether the results of the perceptual experiments correlate with their analytical counterparts. It is found that, when a minimal feature set is used for the HTS context labels, the addition of POS tags does improve the naturalness of the voice. However, the same effect can be accomplished by including segmental counting and positional information instead of the POS tags. / Thesis (M.Sc. Engineering Sciences (Electrical and Electronic Engineering))--North-West University, Potchefstroom Campus, 2011.

In the world of human language technology, resource–scarce languages (RSLs) suffer from the problem of little available electronic data and linguistic expertise. The Lwazi project in South Africa is a large–scale endeavour to collect and apply such resources for all eleven of the official South African languages. One of the deliverables of the project is more natural text–to–speech (TTS) voices. Naturalness is primarily determined by prosody and it is shown that many aspects of prosodic modelling is, in turn, dependent on part–of–speech (POS) information. Solving the POS problem is, therefore, a prudent first step towards meeting the goal of natural TTS voices. In a resource–scarce environment, obtaining and applying the POS information are not trivial. Firstly, an automatic tagger is required to tag the text to be synthesised with POS categories, but state–of–the–art POS taggers are data–driven and thus require large amounts of labelled training data. Secondly, the subsequent processes in TTS that are used to apply the POS information towards prosodic modelling are resource–intensive themselves: some require non–trivial linguistic knowledge; others require labelled data as well. The first problem asks the question of which available POS tagging algorithm will be the most accurate on little training data. This research sets out to answer the question by reviewing the most popular supervised data–driven algorithms. Since literature to date consists mostly of isolated papers discussing one algorithm, the aim of the review is to consolidate the research into a single point of reference. A subsequent experimental investigation compares the tagging algorithms on small training data sets of English and Afrikaans, and it is shown that the hidden Markov model (HMM) tagger outperforms the rest when using both a comprehensive and a reduced POS tagset. Regarding the second problem, the question arises whether it is perhaps possible to circumvent the traditional approaches to prosodic modelling by learning the latter directly from the speech data using POS information. In other words, does the addition of POS features to the HTS context labels improve the naturalness of a TTS voice? Towards answering this question, HTS voices are trained from English and Afrikaans prosodically rich speech. The voices are compared with and without POS features incorporated into the HTS context labels, analytically and perceptually. For the analytical experiments, measures of prosody to quantify the comparisons are explored. It is then also noted whether the results of the perceptual experiments correlate with their analytical counterparts. It is found that, when a minimal feature set is used for the HTS context labels, the addition of POS tags does improve the naturalness of the voice. However, the same effect can be accomplished by including segmental counting and positional information instead of the POS tags. / Thesis (M.Sc. Engineering Sciences (Electrical and Electronic Engineering))--North-West University, Potchefstroom Campus, 2011.

Conservation in Canada is increasingly driven by land use planning processes. Approaches to governing nature conservation have shifted dramatically from protecting isolated pristine areas to greater attention to the remaining fragments of greenspace in urban, semi-urban and rural areas. The ways that societies govern and use nature are always changing, and these physical management actions are connected to deeply rooted cultural norms and values about the ideal relationship between humans and nature. In the land use planning approach to conservation, citizens and governments find value and construct meaning for remaining nature rather than beginning with normative considerations of what is most worthy of protection. At the root of this conservation planning trend is a growing appreciation for hybrid nature that is valued as natural in spite of the past or present influences upon it. This represents a dramatic shift from the traditional values involved in North American nature conservation, where nature was most valued for its perceived separation from human influence and protected to maintain its untouched qualities. In light of these ideological shifts in the ways that Canadians value and in turn manage nature, is there a corresponding change in the ways that conservation activists perceive environmental value and evaluate naturalness? An increasing number of studies demonstrate that public valuation of nature is not limited to pristine environments: even highly disturbed environments can be valued as natural and are not perceived as a form of lesser nature. Conceptions of what is natural and what is not are highly subjective and variable; in particular, the body of work on the social dimensions of both invasive species and ecological restoration demonstrates the ways in which people construct naturalness in accord with their values and cultural context. By exploring the extent to which people perceive invasive species as reducing naturalness and how ecological restoration is perceived to restore it, these subjects serve as excellent conceptual lenses for exploring constructions of nature. This study explores the subtle variations in environmental values and perception of naturalness among a study population who self-identify as pursuing the same goal: ensuring the continued protection of the Oak Ridges Moraine. The Moraine is a partly urbanized landform in southern Ontario that is situated within a complex hybrid socio-ecological landscape. It is also the subject of an active and high profile conservation movement that has spanned over 40 years. Using a combination of interviews and Q Method, this study explored how citizens engaged in Oak Ridges Moraine conservation perceive both the current and ideal state of naturalness on the Moraine, with specific emphasis on how the discourses these citizens use to frame the Moraine invoke the concept of naturalness Findings from this study reveal that Moraine activists represent a conservation paradox: they value the natural, non-human qualities of the landform, yet at the same time identify the Moraine as a hybrid landscape with both social and ecological qualities. In particular, respondents indicated a strong interest in naturalness in the context of invasive species and ecological restoration, yet at the same time identified the naturalness of the Moraine to be a lesser priority in the face of urban development pressures. In this way, citizens engaged in Moraine conservation respond to the hybrid quality of the Moraine landscape by moving beyond the binary distinction between nature and society, situating themselves as both apart from and a part of the landscape at the same time. This finding demonstrates how values for conserving nature are affected by hybridity between social and ecological systems, and suggests how embracing the paradox of hybrid nature can contribute to understanding and managing complex socio-ecological systems.

The proportion of people over 65+ is expected to increase in the future. The elderly is one of the most frequent users of open public space. One type of open public space is urban green space, which is a recurring appreciated element for the elderly population. Despite this, cities are often considered to neglect the needs and wants of the elderly population regarding design and shape. What we need and want in urban green areas differs among different age- and culture groups. Therefor it is important to shed light upon elderly’s experience linked to the use and needs of urban green space to get a better understanding of how these areas are perceived. Which this study aims to do based upon elderly living the city of Umeå. The paper takes a qualitative approach to answer the research questions. Semi-structured interviews were conducted with 7 informants living in the Umeå urban area. Results show that many of the elderly informant’s values urban green space with a high perceived naturalness which is also considered to be lacking in many Umeås central parks. However, what types of greenery which contributes to the perceived naturalness can vary. Woodlands as well as smaller leafy and heavily green influenced areas can equally result in a high perceived naturalness. Areas with a high perceived naturalness also enables elderly’s motives for being in areas of greenery to a greater extent. Like the daily walk, relaxing and socializing with friends. All open green space in the city is not sought-after by the elderly informants. Further factors that make some of the elderly informants seek out different open green areas are lack of amenities and the feeling of inaccessibility.

Submitted by Caroline Periotto (carol@ufscar.br) on 2016-09-12T15:12:07Z No. of bitstreams: 1 TeseIMMS.pdf: 4785300 bytes, checksum: cde00814828a088da4a0c54a4d6558e7 (MD5) / Approved for entry into archive by Marina Freitas (marinapf@ufscar.br) on 2016-09-13T14:30:56Z (GMT) No. of bitstreams: 1 TeseIMMS.pdf: 4785300 bytes, checksum: cde00814828a088da4a0c54a4d6558e7 (MD5) / Approved for entry into archive by Marina Freitas (marinapf@ufscar.br) on 2016-09-13T14:31:04Z (GMT) No. of bitstreams: 1 TeseIMMS.pdf: 4785300 bytes, checksum: cde00814828a088da4a0c54a4d6558e7 (MD5) / Made available in DSpace on 2016-09-13T14:31:13Z (GMT). No. of bitstreams: 1 TeseIMMS.pdf: 4785300 bytes, checksum: cde00814828a088da4a0c54a4d6558e7 (MD5) Previous issue date: 2015-11-30 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) / The complexity and interaction of environmental problems have directed the landscape structure studies, which emphasizes broad spatial scales related to ecological effects of spatial patterns of the ecosystems. Changes in the landscape result in the impairment of natural capital and ecosystem functions that regulate or modulate the goods and services provided to human wellbeing, supported by the properties and/or physical, biological or chemical processes that occur in natural systems. For understanding the trajectories of natural and anthropogenic processes of landscape, many studies have addressed the models based on land use and land cover dynamics, which reflect the potential of the landscape transformation and thus assess implications for environmental sustainability scenario to over time. Thus, this thesis aims to draw up scenarios and projections of landscape patterns resulting from the interaction society and nature, based on structural indicators of the municipality of San Carlos landscape, basing the discussion on changes in natural capital in sustainable capacity in provide ecosystem services. / A complexidade e a interação dos problemas ambientais têm direcionado os estudos de estrutura da paisagem enfatizando escalas espaciais amplas, relacionados aos efeitos ecológicos dos padrões espaciais dos ecossistemas. As alterações na paisagem resultam no comprometimento do capital natural e das funções ecossistêmicas que regulam ou modulam os bens e serviços proporcionados ao bem-estar humano, suportados pelas propriedades e/ou processos físicos, biológicos ou químicos que ocorrem nos sistemas naturais. Para a compreensão das trajetórias dos processos naturais e antrópicos da paisagem, muitos estudos têm abordado os modelos baseados na dinâmica dos usos e cobertura da terra, que permitem evidenciar os possíveis caminhos da transformação da paisagem e dessa forma avaliar implicações no cenário de sustentabilidade ambiental ao longo do tempo. Dessa forma, esta tese pretende a elaboração de cenários e projeções dos padrões da paisagem resultantes da interação sociedade-natureza, com base em indicadores estruturais da paisagem do Município de São Carlos, embasando a discussão sobre as alterações do capital natural na capacidade de sustentabilidade em proporcionar serviços ecossistêmicos.

No description available.

Les premières utilisations des forêts par l’homme remontent au paléolithique ; depuis l’homme n’a cessé de modifier le couvert forestier. Après une introduction sur l’histoire de l’usage anthropique des forêts, trois questions de recherches sont posées : 1- Quelle est la composition des paysages forestiers ? 2- Comment se répartit le paysage végétal forestier ? 3- Comment peut-on évaluer l’impact des interventions humaines sur le paysage végétal forestier ? Le territoire d’étude est le Massif armoricain, territoire le moins boisé de France avec de grandes surfaces agricoles. Cette recherche s’inscrit dans le programme national de cartographie des végétations naturelles et semi-naturelles de France (CarHAB) initié en 2010 par le Ministère de la Transition écologique et solidaire, suivant une approche dynamico-caténale. La symphytosociologie et géosymphytosociologie s’intéressent à la fois aux aspects dynamiques (séries) et à l’intégration spatiale des communautés à différentes échelles (géoséries), et sont donc particulièrement adaptées à l’étude des écocomplexes forestiers. Une typologie des séries et des géoséries de végétation forestière du Massif armoricain est proposée. Une cartographie de la végétation actuelle (VA) et de la végétation potentielle naturelle (VPN) est réalisée sur des sites représentatifs du territoire d’étude. Enfin, une étude diachronique sur le massif d’Andaine (61) permet d’évaluer l’impact des activités humaines passées, actuelles, et de simuler l’évolution future du massif forestier. Ces informations, couplées à la carte de la VPN du site, permettent, de formuler des préconisations de gestion et de fournir aux gestionnaires un outil pour se rapprocher de la VPN, tout en conciliant une exploitation multifonctionnelle du site. / The first human uses of forests date back to the Paleolithic period. Since then, man has constantly modified the forest cover. Based on the history of human forest use, three research questions emerge: 1- what is the forest landscapes composition? 2- How is the forest vegetation landscape distributed? 3- How the human impact on the forest vegetation landscape could be assessed?To answer these questions, the armorican Massif, the most least-forested territory in France with large agricultural areas, is studied. This project is part of the CarHAB national program initiated by the FrenchEcological Ministry in 2010, based on a dynamiccatenal phytosociology approach. Symphytosociology and geosymphytosociology focus on both dynamical aspects (series) and spatial distribution of communities at different scales (geoseries), and are consequently particularly well-adapted for the forest ecocomplexes study. First, a typology of the forest vegetation series and geoseries in the armorican Massif is proposed. Then, a mapping of actual vegetation (VA) and potential natural vegetation (PNV) is done on study area representative sites.Finally, a diachronic study on the Andaine massif (61) is used to assess past and present human impact and to simulate the future evolution of the forest vegetation landscape. Those informations, combined with the PNV site map, are used to develop management guidelines and provide to forest managers a tool to get closer to the PNV while ensuring a multifunctional site operation at the same time.

Este estudo se baseou em dados extraídos de um corpus paralelo composto por textos originais em inglês de artigos da Revista National Geographic, suas respectivas traduções publicadas em português na National Geographic Brasil, e os textos traduzidos antes da revisão para publicação. O objetivo foi cotejar os três textos a fim de identificar, na tradução, elementos que promovem a produção de um texto natural e fluente, que reflete o uso mais corrente na língua. Sua inserção no âmbito da lingüística de corpus se deu por essa permitir um estudo lingüístico descritivo e não prescritivo da língua e por se voltar não só para as palavras, estruturas ou usos possíveis na língua mas também o que é provável que ocorra, isto é, a diferença entre o que os falantes podem dizer e o que na verdade dizem. Esse cotejo possibilitou o levantamento de aspectos lingüísticos que afetam a naturalidade e fluência do texto, como a fraseologia típica na língua de chegada, suas colocações naturais, suas próprias expressões fixas, que refletem um modo preferido de se expressar de cada comunidade lingüística. Esses aspectos estão intimamente relacionados à convencionalidade, que abrange aquilo que é consolidado pelo uso, constituindo muitas vezes fonte de dificuldade ao \"tradutor ingênuo\" que pode não perceber que dentre formas possíveis existe aquela mais provável de ocorrer. Dentre os elementos levantados, selecionou-se a tradução dos advérbios terminados em *ly em inglês, nem sempre traduzidos por advérbios terminados em *mente em português, procurando compreender essa opção dentro do âmbito da convencionalidade. Essas opções nem sempre são ilustradas em dicionários, constituindo a pesquisa baseada em corpora, tanto paralelos quanto monolíngües, um recurso valioso ao tradutor seja para mostrar estratégias empregadas por outros tradutores profissionais como para conscientizá-lo de um uso consagrado na língua. / This study was based on data collected from a parallel corpus composed of original texts in English of articles from National Geographic magazine, and their respective translations in Portuguese as published in National Geographic Brazil and the texts translated before their revision for publication. The aim was to compare the three texts in order to identify, in translation, elements that enable the production of a text which sounds natural and fluent, reflecting the current use by the speakers. Corpus Linguistics was chosen as an approach for the research since it allows a descriptive rather than a prescriptive study of the language and is typically concerned not only with what words, structures or uses are possible in a language, but also with what is probable, likely to occur, the difference between what the speakers can say and what they actually say. This comparison brought to light linguistic features which might affect naturalness and fluency of a text, as the typical phraseology in the target-language, its natural collocations and its own fixed expressions, which reflect a preferred way of a community of saying things. These aspects are closely related to conventionality, which encompasses that which is attested by the use, and represent precisely where the difficulties lie for the \"innocent translator\", who might not notice that among a range of possible forms there is one which is more likely to occur. Among the elements found, one was particularly investigated, namely, the translation of adverbs ending in *ly in English , not always translated by adverbs ending in *mente in Portuguese, with a view to understand this option within the idea of conventionality. These options are not frequently displayed in dictionaries, which makes corpus-based research , either using parallel corpora or monolingual corpora, an invaluable resource to translators, in eliciting strategies employed by other professional translators and in bringing awareness to an attested use in language.

The acquisition of phonological alternations consists of many aspects as discussions in the relevant literature show. There are contrary findings about the role of naturalness. A natural process is grounded in phonetics; they are easy to learn, even in second language acquisition when adults have to learn certain processes that do not occur in their native language. There is also evidence that unnatural – arbitrary – rules can be learned. Current work on the acquisition of morphophonemic alternations suggests that their probability of occurrence is a crucial factor in acquisition. I have conducted an experiment to investigate the effects of naturalness as well as of probability of occurrence with 80 adult native speakers of German. It uses the Artificial Grammar paradigm: Two artificial languages were constructed, each with a particular alternation. In one language the alternation is natural (vowel harmony); in the other language the alternation is arbitrary (a vowel alternation depends on the sonorancy of the first consonant of the stem). The participants were divided in two groups, one group listened to the natural alternation and the other group listened to the unnatural alternation. Each group was divided into two subgroups. One subgroup then was presented with material in which the alternation occurred frequently and the other subgroup was presented with material in which the alternation occurred infrequently. After this exposure phase every participant was asked to produce new words during the test phase. Knowledge about the language-specific alternation pattern was needed to produce the forms correctly as the phonological contexts demanded certain alternants. The group performances have been compared with respect to the effects of naturalness and probability of occurrence. The natural rule was learned more easily than the unnatural one. Frequently presented rules were not learned more easily than the ones that were presented less frequently. Moreover, participants did not learn the unnatural rule at all, whether this rule was presented frequently or infrequently did not matter. There was a tendency that the natural rule was learned more easily if presented frequently than if presented infrequently, but it was not significant due to variability across participants. / Suffixe, die an Wortstämme angehängt werden, tragen grammatische Informationen. Bei Verben wird dabei die Person, Numerus, Tempus, Modus und Genus Verbi angezeigt, bei Nomen Kasus, Numerus und Genus. Durch phonologische Kontexte bedingt kann eine solche morphologische Markierung ihre Gestalt ändern und unterschiedliche Oberflächenformen annehmen. Die dabei entstandenen Allomorphe werden durch regelbasierte Prozesse von dem zugrunde liegenden Morphem abgeleitet. Es zeigt sich, dass der Erwerb morphophonemischer Alternationen ein aufwendiger und schwieriger Lernprozess ist. Die vorliegende Arbeit beschäftigt sich nun mit Faktoren, die den Erwerb der Alternationen positiv beeinflussen können. Zum einen wird der Faktor Natürlichkeit, zum anderen der Faktor Auftretenshäufigkeit diskutiert. Trotz einiger widersprüchlicher Evidenzen bezüglich des ersten Faktors hat sich in der neueren Forschung herausgestellt, dass ein natürlicher Prozess leichter zu lernen ist als ein unnatürlicher. Oft konnte ein Vorteil der natürlichen gegenüber den unnatürlichen Prozessen festgestellt werden. Allerdings zeigt sich dieser Umstand nicht immer – dann wiederum zeigt sich kein Vorteil gegenüber einem der beiden Prozesse. Die Ursachen dafür sind in der Methode oder der Herangehensweise zu suchen. Mache Methode scheint nicht sensitiv genug zu sein, den Vorteil aufzudecken, und manche Studien gehen unterschiedlich an die generelle Frage heran, was denn überhaupt ein natürlicher Prozess ist. Unter Berücksichtigung der einschlägigen Literatur habe ich Charakteristika eines typisch natürlichen Prozesses herausgearbeitet und damit die definitorische Grundlage für die empirische Untersuchung derselben bestimmt. Die Auftretenshäufigkeit eines Prozesses scheint auch ein entscheidender Faktor für den Erwerbsprozess zu sein. Dabei wird der Prozess leichter gelernt, der frequent im Input vorliegt, wohingegen ein Prozess schwieriger zu lernen ist, je weniger häufig er vorkommt. In verschiedenen Studien konnte gezeigt werden, dass die bloße Verteilung eines Musters in Wörtern bzw. im Lexikon schon ausreichen kann, zugrunde liegende Repräsentationen zu formen. Dabei ist immer das frequentere Muster das zuverlässigere. Anhand einer experimentellen Studie habe ich beide Faktoren direkt miteinander verglichen. Es wurde die Lernbarkeit einer natürlichen künstlichen Sprache mit der einer unnatürlichen künstlichen Sprache verglichen. Die Sprachen unterschieden sich lediglich in der Regel, nach der eines von zwei Pluralmorphemen ausgewählt werden musste, wobei die natürliche Sprache nach Vokalharmonie alternierte und die unnatürliche Sprache nach einer arbiträren Regel. In zwei Hauptgruppen wurde 80 erwachsenen Deutschen entweder die eine oder die andere Sprache präsentiert. In jeder Gruppe wurde die Hälfte der Probanden häufig (zu 50%) mit der Alternation konfrontiert, die andere Hälfte infrequent (zu 25%). Nach der Familiarisierungsphase ohne expliziten Lernauftrag war die Aufgabe aller Probanden, von neuen Wörtern der Sprache(n) den Plural zu bilden. Die Analyse der Reaktionen ergab einen Effekt der Natürlichkeit, aber keinen der Auftretenshäufigkeit: Die natürliche Sprache war deutlich besser zu lernen als die unnatürliche. Die Auftretenshäufigkeit in beiden Sprachen führte zu keinem signifikanten Unterschied. Kein einziger von den 40 Probanden, die die unnatürliche Regel präsentiert bekamen, hat die Regel für die entsprechende Alternation gelernt. Es zeigt sich jedoch eine Tendenz bei den Probanden, die die natürliche Sprache erlernen sollten: Diejenigen scheinen einen Vorteil zu haben, die häufiger die Alternation während der Familiarisierungsphase hören. Aber auch unter den Probanden, die mit der natürlichen Sprache konfrontiert wurden, zeigten einige gar keinen Lernerfolg, weshalb ich vermute, dass wegen der großen Varianz in den Daten die Auftretenshäufigkeit als einflussreicher Faktor empirisch nicht belegt werden konnte. Zusammenfassend konnte nachgewiesen werden, dass das Lernen der Alternationen sehr stark von einem bias für Natürlichkeit beeinflusst wird. Allein mit der distributionellen Analyse der verschiedenen Pluralendungen hätte der Vorteil für die Alternation der natürlichen Sprache nicht erklärt werden können.

Tone mapping operators aim to compress high dynamic range (HDR) images to low dynamic range ones so as to visualize HDR images on standard displays. Most existing works were demonstrated on specific examples without being thoroughly tested on well-established and subject-validated image quality assessment models. A recent tone mapped image quality index (TMQI) made the first attempt on objective quality assessment of tone mapped images. TMQI consists of two fundamental building blocks: structural fidelity and statistical naturalness. In this thesis, we propose an enhanced tone mapped image quality index (eTMQI) by 1) constructing an improved nonlinear mapping function to better account for the local contrast visibility of HDR images and 2) developing an image dependent statistical naturalness model to quantify the unnaturalness of tone mapped images based on a subjective study. Experiments show that the modified structural fidelity and statistical naturalness terms in eTMQI better correlate with subjective quality evaluations. Furthermore, we propose an iterative optimization algorithm for tone mapping. The advantages of this algorithm are twofold: 1) eTMQI and TMQI can be compared in a more straightforward way; 2) better quality tone mapped images can be automatically generated by using eTMQI as the optimization goal. Numerical and subjective experiments demonstrate that eTMQI is a superior objective quality assessment metric for tone mapped images and consistently outperforms TMQI.