Dissertations / Theses on the topic 'Subtitle analysis'

As film-making has evolved, subtitling has become “a translation practice that consists of presenting a written text, generally on the lower part of the screen”, which seeks to deliver “the original dialogue of the speakers” (Diaz-Cintas and Remael 2007: 8). Subtitle translation in the field of Audiovisual Translation has attracted increasing interest from scholars, who believe that the quality of subtitling is an important criterion when evaluating the quality of a film. Inspired by Even-Zohar’s (1978) polysystem theory and Gottlieb’s (1992) typology, this comparative study focuses on the subtitling of English-language animation films, which represent one of the main foreign film genres introduced to Chinese audiences, in order to identify effective strategies to improve the quality of translated subtitles from the perspective of cultural awareness. I propose three main hypotheses in this research: 1) that the length of a film subtitle translated into Chinese is generally greater than the corresponding original English text; 2) that effective translation is contingent on the use of Chinese idioms and traditional expressions within translated subtitles; 3) that effective translation is contingent on the use of popular expressions within contemporary Chinese culture. The methodologies employed to address these hypotheses are: a self-built parallel corpus (as the main research tool) comprising 15 sets of animation film subtitles; a questionnaire survey; and interviews with relevant AVT practitioners and scholars. The research findings are: 1) the length of a film subtitle translated into Chinese is generally greater than the corresponding original English text; 2) the two discussed types of expression are effectively employed in the data; 3) unnecessary, erroneous or inappropriate uses, among others, were highlighted by the data; 4) the frequency of using popular expressions is higher than for traditional expressions; 5) there is frequent use of the typology strategies of paraphrase, transfer, condensation and transcription, as well as newer strategies of globalization and localization.

The area of audiovisual translation indisputably awakens a growing interest amongst scholars. However, the study of the translation of songs in an audiovisual context (AVC) has not been thoroughly explored. The purpose of this research is to offer a descriptive analysis on the area of translating songs (from English into Spanish) in an AVC. In particular, the research focuses on the translation of the songs included in nine animated Disney films from the Disney Renaissance Period (1989-1999). The study assesses how non-linguistic elements such as rhyme, rhythm, images or lip synchrony affect the task of translating songs by observing specific practical examples. Both subtitling and dubbing are analysed as translation practices. The different constraints involving each form of translation are assessed. Furthermore, the thesis offers a descriptive analysis of strategies used in the translation of songs in AVCs. This study highlights the importance of assessing the impact of non-linguistic elements in future studies on the translation of vocal music in AVCs and aims to provide a model for the contrastive analysis of song lyrics between the ST and the TT.

This thesis examines the subtitling of films depicting the French banlieue into English. The banlieues are housing estates situated on the outskirts of large towns and cities, and are primarily home to the underprivileged, and immigrants to France or their descendants. The sociolect spoken in the banlieue differs from standard French in terms of grammar, lexicon and pronunciation. Three films released between 2000 and the present day are studied; La squale (Genestal, 2000), L'esquive (Kechiche, 2003) and Divines (Benyamina, 2016). A new integrated methodology is developed, which examines the films within their broader contexts of release, and in light of paratextual material contributing to the context of reception, and to the viewer's understanding of the topic at hand. Directors representing the banlieue on screen generally do so with a view to provoking thought or public discussion in relation to the banlieues. In addition to macro- and micro-contextual analysis of the films and subtitles, the work is underpinned by an examination of the subtitling situation, encompassing the views and experiences of subtitlers working on banlieue film, and technical analysis of the subtitles in terms of readability. Through interviews of professional subtitlers, and close technical analysis of the subtitles, this research is contextualised within the industry, and within current conventions and guidelines. Close analysis of subtitles and the translation solutions they present reveals that some of the socio-political messages presented in the films may not be evident to a non-French speaking viewer of the English-subtitled versions. Although the informal nature of many conversations featuring the langage de banlieue is sometimes clear in the subtitled version, the unique sociolect of the characters is not. In two of the case study films, a dialect-for-dialect approach was adopted, where African American vernacular English was used in the subtitles to demonstrate the use of non-standard language. However, it is argued that ultimately, this dialect-for-dialect approach, combined with cultural similarities between the French banlieue and American street culture, could lead the British Anglophone viewer to negotiate the banlieues and those who live there via their knowledge of American street culture. This could contribute to American cultural hegemony, and does not convey the specificity of France's banlieues as cultural melting pots.

The purpose of this thesis is to explore how the approaches identified for the translation of Taiwanese Mandarin dialogues into English subtitles is different from those described in research studies which examined the translation of English film dialogues into Mandarin subtitles using data gathered from a Taiwanese movie entitled Cape No. 7. Taiwanese Mandarin is significantly influenced by the Tai-yu dialect, and in order to carry out the translation of Taiwanese Mandarin, it is important to deal with the functions of the dialect and related culture-specific items (CSIs). Therefore, the researcher of this thesis has investigated whether the translator of Cape No. 7 successfully applied a variety of translation strategies and prioritized the translation criteria well, to enable the target audience to receive the same message of the film as the source audience, especially when subtitling is constrained by time and space. There are two research areas which have been examined: slang and humour. The data analysis of slang scenes included Tai-yu slang and swearing, while humorous scenes were divided into three areas: puns, irony and metaphor. The research data have been analysed by means of analysis diagrams, which were trialled in a pilot study, described in Chapter Three. The pilot study lead to a new categorization based on previous researchers’ taxonomies (Aixelá, 1996; Davies, 2003; Tveit and Fong 2005, as cited in Yang, 2006), and this was applied to the analysis diagrams. According to the findings from the analysis chapters (Chapter Four & Five), the translator of Cape No. 7 preferred to use mostly the paraphrasing strategy in translating Tai-yu slang and humour, and the synonymy strategy in conveying swearing. In addition, the translator often strengthened swear words rather than toning them down in the English subtitles. However, when subtitling humorous scenes, the translator preferred to incorporate two strategies to deal with the cultural barriers when conveying the humorous effect. The compensation strategy was not used often in slang translation. In regard to the translation criteria, pragmatics and accuracy appeared to have been deemed more important by the translator than other criteria. Surprisingly functional equivalence did not seem to have been the translator’s first priority, as suggested by earlier research. Overall, this research study appeared to show that the translator of Cape No. 7 considered the paraphrase strategy to be the most efficient strategy for maintaining CSIs, and that pragmatics and accuracy were the translator’s most important criteria, which differs from the findings of previous research studies.

The aim of the study is to apply a paratext analysis to investigate what it is in thepackaging of books that can appear attractive to children and what are the distinctiveshared features of popular pupil choices among books. The analysis consists of anexamination of each book and then a comparative section. The analytical categories arethe title and subtitle, author’s name, publisher’s information, cover pictures, the backcover, the spine, the format of the book, endpapers, and the series emblem. Thetheoretical premises are hermeneutics and semiotic interpretations of signs. The studyproceeds from Genette’s concepts of paratext, peritext and epitext. A prominent common feature of the books was series emblems in the form ofmain titles and colours. Another significant conclusion is that the cover often depictssome climactic event in the content of the book. The content and the cover mustharmonize so that the reader will not feel disappointed. The didactic conclusions are thatteachers must have a good understanding of the effect a book cover can have on pupils.Often teachers try to motivate pupils by telling them about the exciting content, but ifthe pupil is not motivated by the paratext, the threshold will be too high. It is of greatrelevance that teachers offer pupils several different books with varying paratexts.

Dans le contexte en constante évolution du contenu audiovisuel, la nécessité cruciale d'automatiser l'indexation et l'organisation des archives s'est imposée comme un objectif primordial. En réponse, cette recherche explore l'utilisation de techniques d'apprentissage profond pour automatiser l'extraction de métadonnées diverses dans les archives, améliorant ainsi leur accessibilité et leur réutilisation. La première contribution de cette recherche concerne la classification des mouvements de caméra. Il s'agit d'un aspect crucial de l'indexation du contenu, car il permet une catégorisation efficace et une récupération du contenu vidéo en fonction de la dynamique visuelle qu'il présente. L'approche proposée utilise des réseaux neuronaux convolutionnels 3D avec des blocs résiduels. Une approche semi-automatique pour la construction d'un ensemble de données fiable sur les mouvements de caméra à partir de vidéos disponibles au public est également présentée, réduisant au minimum le besoin d'intervention manuelle. De plus, la création d'un ensemble de données d'évaluation exigeant, comprenant des vidéos de la vie réelle tournées avec des caméras professionnelles à différentes résolutions, met en évidence la robustesse et la capacité de généralisation de la technique proposée, atteignant un taux de précision moyen de 94 %.La deuxième contribution se concentre sur la tâche de Vidéo Question Answering. Dans ce contexte, notre Framework intègre un Transformers léger et un module de cross modalité. Ce module utilise une corrélation croisée pour permettre un apprentissage réciproque entre les caractéristiques visuelles conditionnées par le texte et les caractéristiques textuelles conditionnées par la vidéo. De plus, un scénario de test adversarial avec des questions reformulées met en évidence la robustesse du modèle et son applicabilité dans le monde réel. Les résultats expérimentaux sur MSVD-QA et MSRVTT-QA, valident la méthodologie proposée, avec une précision moyenne de 45 % et 42 % respectivement. La troisième contribution de cette recherche aborde le problème de vidéo captioning. Le travail introduit intègre un module de modality attention qui capture les relations complexes entre les données visuelles et textuelles à l'aide d'une corrélation croisée. De plus, l'intégration de l'attention temporelle améliore la capacité du modèle à produire des légendes significatives en tenant compte de la dynamique temporelle du contenu vidéo. Notre travail intègre également une tâche auxiliaire utilisant une fonction de perte contrastive, ce qui favorise la généralisation du modèle et une compréhension plus approfondie des relations intermodales et des sémantiques sous-jacentes. L'utilisation d'une architecture de transformer pour l'encodage et le décodage améliore considérablement la capacité du modèle à capturer les interdépendances entre les données textuelles et vidéo. La recherche valide la méthodologie proposée par une évaluation rigoureuse sur MSRVTT, atteignant des scores BLEU4, ROUGE et METEOR de 0,4408, 0,6291 et 0,3082 respectivement. Notre approche surpasse les méthodes de l'état de l'art, avec des gains de performance allant de 1,21 % à 1,52 % pour les trois métriques considérées. En conclusion, ce manuscrit offre une exploration holistique des techniques basées sur l'apprentissage profond pour automatiser l'indexation du contenu télévisuel, en abordant la nature laborieuse et chronophage de l'indexation manuelle. Les contributions englobent la classification des types de mouvements de caméra, la vidéo question answering et la vidéo captioning, faisant avancer collectivement l'état de l'art et fournissant des informations précieuses pour les chercheurs dans le domaine. Ces découvertes ont non seulement des applications pratiques pour la recherche et l'indexation de contenu, mais contribuent également à l'avancement plus large des méthodologies d'apprentissage profond dans le contexte multimodal
Within the dynamic landscape of television content, the critical need to automate the indexing and organization of archives has emerged as a paramount objective. In response, this research explores the use of deep learning techniques to automate the extraction of diverse metadata from television archives, improving their accessibility and reuse.The first contribution of this research revolves around the classification of camera motion types. This is a crucial aspect of content indexing as it allows for efficient categorization and retrieval of video content based on the visual dynamics it exhibits. The novel approach proposed employs 3D convolutional neural networks with residual blocks, a technique inspired by action recognition methods. A semi-automatic approach for constructing a reliable camera motion dataset from publicly available videos is also presented, minimizing the need for manual intervention. Additionally, the creation of a challenging evaluation dataset, comprising real-life videos shot with professional cameras at varying resolutions, underlines the robustness and generalization power of the proposed technique, achieving an average accuracy rate of 94%.The second contribution centers on the demanding task of Video Question Answering. In this context, we explore the effectiveness of attention-based transformers for facilitating grounded multimodal learning. The challenge here lies in bridging the gap between the visual and textual modalities and mitigating the quadratic complexity of transformer models. To address these issues, a novel framework is introduced, which incorporates a lightweight transformer and a cross-modality module. This module leverages cross-correlation to enable reciprocal learning between text-conditioned visual features and video-conditioned textual features. Furthermore, an adversarial testing scenario with rephrased questions highlights the model's robustness and real-world applicability. Experimental results on benchmark datasets, such as MSVD-QA and MSRVTT-QA, validate the proposed methodology, with an average accuracy of 45% and 42%, respectively, which represents notable improvements over existing approaches.The third contribution of this research addresses the multimodal video captioning problem, a critical aspect of content indexing. The introduced framework incorporates a modality-attention module that captures the intricate relationships between visual and textual data using cross-correlation. Moreover, the integration of temporal attention enhances the model's ability to produce meaningful captions, considering the temporal dynamics of video content. Our work also incorporates an auxiliary task employing a contrastive loss function, which promotes model generalization and a deeper understanding of inter-modal relationships and underlying semantics. The utilization of a transformer architecture for encoding and decoding significantly enhances the model's capacity to capture interdependencies between text and video data. The research validates the proposed methodology through rigorous evaluation on the MSRVTT benchmark,viachieving BLEU4, ROUGE, and METEOR scores of 0.4408, 0.6291 and 0.3082, respectively. In comparison to state-of-the-art methods, this approach consistently outperforms, with performance gains ranging from 1.21% to 1.52% across the three metrics considered.In conclusion, this manuscript offers a holistic exploration of deep learning-based techniques to automate television content indexing, addressing the labor-intensive and time-consuming nature of manual indexing. The contributions encompass camera motion type classification, VideoQA, and multimodal video captioning, collectively advancing the state of the art and providing valuable insights for researchers in the field. These findings not only have practical applications for content retrieval and indexing but also contribute to the broader advancement of deep learning methodologies in the multimodal context

This study is an analysis of the process of creating subtitles for the National Geographic documentary Fight Science: Mixed Martial Arts. The difficulties encountered during this project have included the translation of subject-specific terminology, the handling of source-text mistakes and the process of deciding how to condense the spoken dialogue when writing the subtitles. By studying these issues, certain conclusion can be drawn about the difficulties encountered when creating subtitles and translating texts with subject-specific terminology. The study is primarily based on the theories of Runo Ingo (2007), Vinay & Darbelnet (1977) and Chesterman & Wagner (2002). To find appropriate translations for subject-specific terminology, it is important to understand which translation device should be used (for example a loan or literal translation) through the examination of parallel texts and it is important that the translator be familiar with this terminology. How to properly handle mistakes in the source material depends greatly on the situation, the text concerned and the translator’s judgment. Decisions about which information can be omitted can be based on the information’s importance to the overall understanding of the show or the implicit nature of the information. The science of translation can be considered vague, but there exist theories and principles to aid translators when making such touch decisions as those encountered during this project.

Although Audiovisual Translation (AVT) studies have proliferated in the past two decades, studies that tackle AVT in Arabic are still in their early stages. More specifically, very few studies investigate subtitling as a norm-governed, micro-level process that reflects macro-level, socio-cultural structures. Therefore, this thesis investigates the subtitling of racial slurs from English to Arabic by utilising the principles of Critical Discourse Analysis (CDA) and Descriptive Translation Studies (DTS). It seeks to identify the main subtitling strategies that Arab subtitlers use to translate racial slurs in English-language films into Arabic and the ideological underpinnings behind those strategies. Racial slurs are seen as aspects of lingua-culture; their occurrence represents the ideologies of race and racism that operate within a given socio-culture. This study shows that racial slurs serve as tools for characterisation through stereotyping. Moreover, the study demonstrates how racial slurs are a manifestation of the audience design involved in producing a film’s script. A dataset that consists of 556 racial slurs found in 102 films which were recorded from Arab satellite channels is compiled. The study identifies transfer, decimation, imitation, resignation, and omission as the main strategies that Arab subtitlers use in translating racial slurs. After the main subtitling strategies have been identified, the dataset is approached from the angle of racial categories targeted by racial slurs. This approach forms the basis of the in-depth, qualitative analysis of several representative scenes from the films under investigation. First, the analysis focuses on racial slurs targeting Blacks, which are the most recurring in the dataset. Then I analyse the sporadic racial categories of Whites, Asians, Latinos, and Jews. Finally, I focus on racial slurs that target Arabs and Muslims, as these groups represent the members of the audience in the Arab socio-culture. The analysis shows that the subtitling strategy of transfer is more prone to be used with racial slurs that have readily available equivalents in Arabic, such as those that target Blacks and Jews. Reduction strategies are used more frequently with sporadic racial categories and with those categories in which finding an equivalent in Arabic poses a challenge. I argue that the subtitling of racial slurs represents a process in which slurs are entextualised from their original contexts and their linguistic and lingua-cultural layers thereby altered. This process serves as an index of the different ideologies of racism that exist in Western and Arab socio-cultural contexts. It also manifests the role of patronage in controlling the subtitling industry and the subtitled product that reaches Arab audiences. It can be suggested that Arab subtitlers tend to apply domestication as a governing norm in the subtitling of racial slurs, as the acceptability of the subtitled text for the Arab audience seems to be the main priority for subtitlers.

Le criblage à haut contenu connaît un essor important depuis le milieu des années 2000. Cette technologie est d’un intérêt primordial pour l’industrie pharmaceutique car elle permet en principe la découverte de molécules à visée thérapeutique pour des maladies dont les voies moléculaires sont mal identifiées. Jusqu’à présent un phénotype cellulaire mesurable doit au préalable être identifié afin d'évaluer l’effet d’une banque de composés sur celui-ci. Du point de vue de l’analyse d’image, les cellules traitées sont détectée automatiquement sur des centaines de milliers d’images et des mesures de descripteurs permettent de différencier finement les traitements effectifs par rapport à un contrôle négatif. En collaboration avec les entreprises Ksilink et Sanofi, nous avons été confrontés à un nouveau type de crible à haut contenu pour identifier des composés efficaces contre la maladie de Parkinson. Celui-ci, présentant des images de neurones, rendait caduque l’analyse dépendant d’une segmentation robuste des cellules. Par ailleurs, grâce à des techniques de différentiation cellulaire et d’édition de génome, les contrôles du modèle cellulaire sont deux cultures de neurones, isogéniques à une mutation près : la mutation GS2019 induisant la maladie. Néanmoins, l’hétérogénéité propre à ces cellules de forme complexe, et les différences fines entre les deux lignées isogéniques ne permettent pas à l’œil humain l’identification de deux phénotypes distincts. Dans le but de permettre la mise en évidence automatique des différences entre phénotypes, nous avons proposé d’employer des approches de deep learning. Ce travail s’est décomposé principalement en deux étapes. La première étape a consisté à identifier une architecture de réseau capable de classifier des images de neurones. Nous avons appris que les cultures de neurones montraient des différences de phénotype de manière très hétérogène et donc pas systématique. La deuxième étape a consisté à proposer des méthodes pour expliquer et interpréter les différences subtiles de phénotype, ceci afin de s’assurer que le crible soit effectué sur la base d’une différence avérée entre phénotypes et non sur la base d’un biais technique. Partant du principe que les différences entre phénotypes neuronaux sont difficiles à appréhender visuellement entre deux images différentes du à la grande variabilité naturelle des neurones, nous proposons l’idée que la transformation d’une même image d’un phénotype à un autre peut représenter une approche intéressante. En effet, nous montrons qu'il est possible d'entraîner des réseaux antagonistes à transformer une image de neurone non porteur de la mutation en neurone porteur et inversement. De cette façon, nous avons pu mettre en évidence des potentiels biais de l’assay mais également ce que nous pensons être de véritables différences morphologiques liés à la mutation pathologique auparavant invisibles
High-content screening has experienced a significant growth since the mid-2000s. This technology is of primary interest to the pharmaceutical industry as it allows in principle the discovery of therapeutic molecules for diseases whose molecular pathways are poorly identified. Until now, a measurable cell phenotype must first be identified in order to evaluate the effect of a compound library on it. From an image analysis point of view, treated cells are automatically detected in hundreds of thousands of images and measurements of descriptors allow to finely differentiate effective treatments from a negative control. In collaboration with the companies Ksilink and Sanofi, we have been confronted with a new type of high-content screen to identify compounds effective against Parkinson's disease. This one, presenting images of neurons, made analysis dependent on robust cell segmentation obsolete. In addition, using cell differentiation and genome editing techniques, the controls of the cell model are two neuron cultures, isogenic but for one mutation: the GS2019 mutation that induces the disease. Nevertheless, the heterogeneity of these complex cells and the fine differences between the two isogenic lines do not allow the human eye to identify two distinct phenotypes. In order to allow the automatic detection of differences between phenotypes, we have proposed to use deep learning approaches. This work was mainly divided into two steps. The first step consisted in identifying a network architecture capable of classifying neural images. We learned that neuron cultures show phenotype differences in a very heterogeneous and therefore not systematic way. The second step consisted in proposing methods to explain and interpret subtle differences in phenotype, to ensure that the screening is performed on the basis of a proven difference between phenotypes and not on the basis of a technical bias. Based on the premise that differences between neuronal phenotypes are difficult to visually apprehend between two different images due to the high natural variability of neurons, we propose the idea that transforming the same image from one phenotype to another may represent an interesting approach. Indeed, we show that it is possible to train antagonistic networks to transform an image of a neuron that does not carry the mutation into a neuron that carries it and vice versa. In this way, we have been able to highlight potential assay biases but also what we believe to be true morphological differences related to the pathological mutation that were previously invisible

Subtitling for the D/deaf and HoH (SDH) is an invaluable service to millions of D/deaf and HoH viewers providing them with access to audiovisual media broadcast on television and increasingly in cinemas, DVDs and on the Internet. It is also a clear example of an AVT activity where the requirement to know and understand the intended addressee, a requirement present in all types of translation, is all-important. Or is it? Subtitles today still purport to be suitable for all, when by all accounts not only are the D/deaf and HoH two different groups, but what is more important within each group members differ not only in outlook, but also in language and communication needs. The time now seems opportune to answer crucial questions and identify the addressees and their needs/preferences. In order to provide an adequate reply to such questions, a tripartite large-scale research plan is deemed essential. This would combine i) corpus-based research describing current subtitle output with ii) empirical research into the needs and preferences of D/deaf and HoH people and iii) compare both with existing guidelines and update them as necessary.

The aim of this thesis is to investigate the translation discrepancies between the English and the Italian versions of four episodes of the famous sitcom Friends. After reading some previous analysis on Internet, the attention has been focused on some extracts examined mainly in line with the theories of Iaia (2015), Taylor and Perego (2016) and Fusco (2012). The work is divided into three chapters. The first chapter provides a general overview of audiovisual translation with a focus on dubbing, subtitling and humour. The second chapter begins with the plot of Friends and then shows how Friends is currently perceived by the Millennials who have accused it for its lack of diversity and its tendency to popularise homophobia and sexism. The chapter also discusses humour and, in particular, the difficulties of conveying the original humour through audiovisual translation. Finally, its perception in Italy is examined in connection with the role of censorship in the 90s. The third chapter represents the core of this thesis. The focus is on the differences between the English text and the Italian one both in dubbing and subtitling and then compares the two versions to the solutions produced by myself by means of the audiovisual translation program Aegisub. The conclusion will include some considerations about the future of the translation profession and the importance of adapting texts to the current audience where the previous translations are now obsolete.

In the paper we will analyze the Italian localization of the first 5 chapters of the videogame The Last of Us. We will also offer alternative translations and will correct the localization mistakes. The paper consists of five chapters, introduction and conclusion. Each of the chapters represents a category of the revised translation: done badly, omitted translations, done well, could be improved, wrong register or wrong swear words translations. We will observe and analyze the Italian localization of the first five chapters of the videogame and propose the alternatives in the cases of errors and ambiguities, as well as highlight well done translations that reflect the differences between English and Italian grammar, syntax and semantics. The revised parts of the localization primarily include the translation of the subtitles and the artifacts (written documents, not voiced) within the game. The dubbing and the lip sync are not analyzed, unless they influence the translation in one way or another. The orthography and punctuation of the subtitles are kept the same in the thesis as they are in the game. The swear words will be fully spelled for transparency. In the conclusion we will close with the state of the Italian localization of videogames in general and of this one in particular.

Les micro-expressions sont des expressions faciales brèves et subtiles qui apparaissent et disparaissent en une fraction de seconde. Ce type d'expressions reflèterait "l'intention réelle" de l'être humain. Elles ont été étudiées pour mieux comprendre les communications non verbales et dans un contexte médicale lorsqu'il devient presque impossible d'engager une conversation ou d'essayer de traduire les émotions du visage ou le langage corporel d'un patient. Cependant, détecter et reconnaître les micro-expressions est une tâche difficile pour l'homme. Il peut donc être pertinent de développer des systèmes d'aide à la communication exploitant les micro-expressions. De nombreux travaux ont été réalisés dans les domaines de l'informatique affective et de la vision par ordinateur pour analyser les micro-expressions, mais une grande majorité de ces méthodes repose essentiellement sur des méthodes de vision par ordinateur classiques telles que les motifs binaires locaux, les histogrammes de gradients orientés et le flux optique. Étant donné que ce domaine de recherche est relativement nouveau, d'autres pistes restent à explorer. Dans cette thèse, nous présentons une nouvelle méthodologie pour l'analyse des petits mouvements (que nous appellerons par la suite mouvements subtils) et des micro-expressions. Nous proposons d'utiliser la pyramide de Riesz, une approximation multi-échelle et directionnelle de la transformation de Riesz qui a été utilisée pour l'amplification du mouvement dans les vidéos à l'aide de l'estimation de la phase 2D locale. Pour l'étape générale d'analyse de mouvements subtils, nous transformons une séquence d'images avec la pyramide de Riesz, extrayons et filtrons les variations de phase de l'image. Ces variations de phase sont en lien avec le mouvement. De plus, nous isolons les régions d'intérêt où des mouvements subtils pourraient avoir lieu en masquant les zones de bruit à l'aide de l'amplitude locale. La séquence d'image est transformée en un signal ID utilisé pour l'analyse temporelle et la détection de mouvement subtils. Nous avons créé notre propre base de données de séquences de mouvements subtils pour tester notre méthode. Pour l'étape de détection de micro-expressions, nous adaptons la méthode précédente au traitement de certaines régions d'intérêt du visage. Nous développons également une méthode heuristique pour détecter les micro-événements faciaux qui sépare les micro-expressions réelles des clignotements et des mouvements subtils des yeux. Pour la classification des micro-expressions, nous exploitons l'invariance, sur de courtes durées, de l'orientation dominante issue de la transformation de Riesz afin de moyenner la séquence d'une micro-expression en une paire d'images. A partir de ces images, nous définissons le descripteur MORF (Mean Oriented Riesz Feature) constitué d'histogrammes d'orientation. Les performances de nos méthodes sont évaluées à l'aide de deux bases de données de micro-expressions spontanées
Micro-expressions are brief and subtle facial expressions that go on and off the face in a fraction of a second. This kind of facial expressions usually occurs in high stake situations and is considered to reflect a humans real intent. They have been studied to better understand non-verbal communications and in medical applications where is almost impossible to engage in a conversation or try to read the facial emotions or body language of a patient. There has been some interest works in micro-expression analysis, however, a great majority of these methods are based on classically established computer vision methods such as local binary patterns, histogram of gradients and optical flow. Considering the fact that this area of research is relatively new, much contributions remains to be made. ln this thesis, we present a novel methodology for subtle motion and micro-expression analysis. We propose to use the Riesz pyramid, a multi-scale steerable Hilbert transformer which has been used for 2-D phase representation and video amplification, as the basis for our methodology. For the general subtle motion analysis step, we transform an image sequence with the Riesz pyramid, extract and lifter the image phase variations as proxies for motion. Furthermore, we isolate regions of intcrcst where subtle motion might take place and mask noisy areas by thresholding the local amplitude. The total sequence is transformed into a ID signal which is used fo temporal analysis and subtle motion spotting. We create our own database of subtle motion sequences to test our method. For the micro-expression spotting step, we adapt the previous method to process some facial regions of interest. We also develop a heuristic method to detect facial micro-events that separates real micro-expressions from eye blinkings and subtle eye movements. For the micro-expression classification step, we exploit the dominant orientation constancy fom the Riesz transform to average the micro-expression sequence into an image pair. Based on that, we introduce the Mean Oriented Riesz Feature descriptor. The accuracy of our methods are tested in Iwo spontaneous micro-expressions databases. Furthermore, wc analyse the parameter variations and their effect in our results

In recent years, relating gene expression to cancer development and treatment has received a lot of attention. The precise identification of a cancer's type and stage is often crucial to the assignment of appropriate treatment. Therefore, a central goal of the analysis of gene expression data is to identify sets of genes that can be used for the classification or diagnosis of cancer. Another important purpose of gene expression studies is to improve the understanding of cellular responses to drug treatment, to identify drug responsive genes, and to discover the potential drug targets. To achieve these goals, researchers are applying various methods of analysis to gene expression data. Unfortunately, the availability of effective analysis tools lags far behind the availability of data. In this thesis, we present the Gene Expression Analyzer (GEA) for performing cluster analysis on gene expression data. In particular, the GEA is developed to address the reality that cluster analysis is typically a multi-step process. The underlying model of the GEA provides a set of algebraic operators for manipulating the data, as well as intermediate results. Moreover, the GEA provides facilities to help the user to identify candidate genes for further clinical analysis. Last but not least, the GEA is optimized to handle the high dimensionality of gene expression data, and provides user-friendly features to help the user organize and manage intermediate analysis results.

碩士
國立臺灣科技大學
應用外語系
102
For the past years, the film industry worldwide has shot movies and TV series about aviation. And the field of air traffic control under aviation is rarely known to the public. Since the subtitles of the films play an important role, the audience with few background knowledge of aviation even air traffic control will rely on the subtitles to comprehend the plot. The aim of the thesis is to examine and analyze the strategies used in translating the subtitles of films related to aviation and to provide more effective and practicable solutions to the problematic dialogue in the films. The aviation jargon is collected from six English films and one TV series. The suggested translations of problematic subtitles are proposed based on the perspective and what had been learned as an air traffic controller. The result shows that “deletion,” “decimation” and “direct approach” are the strategies frequently used to translate aviation jargon and terms. The quality of the subtitle translations is influenced by the listening ability and professional knowledge of the subtitle translators. The three main factors in mistranslation are caused by the misinterpretation of the nonstandard accents and the images in the films and the lack of specialized knowledge in aviation. It is suggested that subtitle translators apply strategies like “expansion” and “annotation” to clarify the meaning of the terms during the process of translation, and then “condense” the subtitles to fewer words in accordance with the images in films in order to conform to the limited space and help the audience grasp the plot easily and get immersed in the world of aviation with ease.

碩士
國立高雄第一科技大學
應用英語系口筆譯碩士班
105
Crowdsourcing translation is an emerging working model that has gained in popularity in recent years due to the increase in demand for translation services. The benefits of crowdsourcing translation are that it can increase the speed of the process and perhaps save money if translators are volunteers. However, the unstable quality of the texts produced by volunteer translators is one of the main negative effects that has often been noted by professionals. This study aims to investigate the phenomena of crowdsourcing translation by examining the TED Translators. TED Translators offers an example to show how TED refines its crowdsourcing translation process in terms of a review mechanism and subtitling platform. The research method is based on the Descriptive Translation Studies (DTS) approach, as proposed by Toury (1995), and this is used to investigate the norms and phenomena in TED Talk subtitle translation. Guidelines established by TED and existing subtitling conventions are incorporated to investigate the collected TED Talk subtitle translations. The data for this study are collected from TED Talks available on TED.com, translated from English to Chinese, during May 2009 to February 2012, with each talk lasting 12 to 18 minutes. The data analysis shows that TED Talk subtitle translations tend to show five characteristics, which are non-linear translation, lengthy subtitles, improper line breaks, inconsistent punctuation and extra notes, in comparison with conventional subtitle translation. These factors are due to the volunteer translators preserving detailed information, not conforming to the TED guidelines, and the platform lacking certain necessary functions. These characteristics should not be seen as errors comparing with conventional subtitling; however, their distinctive characteristics may consider a norm or phenomena that show volunteer translation’s unique nature. From the analysis of TED Talk subtitle translation and the working process of TED Translators carried out in this work, volunteer translators and subtitling platform are the key elements in contributing to the quality work of TED Talk subtitle translation.

碩士
世新大學
資訊傳播學研究所(含碩專班)
105
The emergence of the online or network video "bullet screen" has facilitated a new comment way. With more and more viewers of using bullet screen, the commentary in the bullet screen and the sentiment expression of the viewer will be more exploratory and value of study. Bullet screen commentary on behalf of the views on the video subjective comments; subtitles is the objective presentation of the role's talked in the video, the relationship between the two is the viewer's view and impressions of the plot in the drama. This study adopts the method of sentiment analysis, carry on statistical analysis to bullet screen and subtitle, compare theirs nature, characteristics and the relationship between the two. The results of the study found that the bullet screen aspect of the word is short, with a strong sentiment, and there are many proprietary emerging vocabularies. Besides, according to data with quantized sentiment, can understand audiences for the degree of likes and dislikes. On the contrary, the subtitle is the basic judgment of the story is good or bad. The analysis of the results of the study can be found closer to the user for the video content broadcast plot of the most immediate and real ideas, can also be applied to the results of the follow-up network video design, recommend for video-related industries, to provide a novel message analysis mode to get the source of the user willingness.

碩士
中華大學
資訊工程學系
106
English is the most common language in the world. Having a good English ability could be helpful in communication and reading, so many countries highly emphasize on English education and teach students English from elementary school. Nowadays, how to learn Engligh well is obviously an important issue. To help students learn English well, teachers spend a lot of efforts in selecting and creating English teaching materials. Although many English e-learning systems have been developed to help students learn English, it is hard to get interesting and royalty free videos as English learning materials. In this thesis, an English e-learning system, called VSAELS, is developed based on the video subtitle analysis technology. VSAELS utilized the online videos on Youtube as the source of learning materials. By analyzing the video subtitles, English vocabularies contained in the videos are automatically extracted for the students to playback and learn. The learning portofolios are also recorded for futher analysis of students’ learning bahaviors. To investigate the impact of VSAELS on English study, a quasi-experiment was conducted in an elementary school English course. The experimental results indicated that VSAELS not only improved the students’ learning achievement, but also promoted their learning motivation, learning attitude, self-efficacy, and flow experience, while the cognitive load was reduced.

碩士
國立臺灣科技大學
應用外語系
101
The boosting growth of foreign films nowadays leads to the importance of subtitling. A film is constructed with dialogues produced by characters with different personalities, personal relationships, social statuses, and gained knowledge. Based on analyzing the language use in discourses, characters’ traits and identities can be clearly reflected. However, no related study has been reported using discourse analysis to investigate language use in both source text and target text. The aim of this research is to explore how discourses represent the language use of characters and realize the goal of the story’s author. Moreover, the researcher intends to assess subtitle translation by looking into whether the language use in the source text is equivalently reproduced. The thesis adopted qualitative approach for the two-stage exploration. First, the researcher analyzed the language use in the source text based on characters’ use of codes, speaking style, and tones. The second stage repeated the same procedure as the first one to analyze the language use in subtitle translations. Afterwards, the researcher assessed the degree of translation’s equivalence and appropriateness and also provided suggested translations. The result showed that mistranslation or over-translation makes the characters’ orientation and image disordered. To equivalently reproduce the effect of source text in the subtitle translation, translators have to rely on knowledge of language use in creating dialogues and pay attention to “what kind of person produces what utterance” and “with whom the person is having conversation”.

Pain is a common experience for children, but there can be tremendous variability in pain expression from one child to another. Clinical lore and psychological theory have highlighted the important role parents may play in socializing children's pain behaviours. For example, certain parental behaviours (e.g., reassurance, criticism) have been linked with increases in child distress, while other behaviours (e.g., humour, suggestions on how to cope) have been linked with decreases in child distress. To date, however, the majority of research examining the relationship between parental behaviours and children's pain experiences has been correlational and conclusions regarding causality have been limited. The purpose of the present study was to provide an experimental investigation of the impact of maternal behaviour on children's pain experiences. Participants were 120 children (60 boys, 60 girls) between the ages of 8 and 12 years and their mothers who were recruited from the community. Mothers were trained to interact with their children in one of three randomly assigned ways during a lab-induced cold pressor pain paradigm: 1) a pain-promoting interaction, consisting of reassurance, empathy, apologies, giving control to the child, and criticism; 2) a pain-reducing interaction, consisting o f non-procedural talk, humour, and suggestions on how to cope; or 3) a no training control group. The content of the maternal training conditions was based on the previous correlational literature indicating links between certain kinds of parental behaviours and child distress. Children's pain experiences were assessed using self-report measures of intensity and affect, behavioural coding of facial expressivity, pain tolerance, and heart rate responsiveness during the cold pressor. Results indicated that maternal interaction type had the expected effects on girls' self-reported ratings of pain intensity; when controlling for the effects of age, girls whose mothers had interacted with them in the pain-promoting manner reported more pain than girls whose mothers were in the control group, who in turn reported more pain than girls whose mothers had interacted with them in the pain-reducing manner. This effect was not significant for boys. Maternal interaction type had no effect on children's facial expressivity, ratings of pain affect, pain tolerance, or heart rate responsiveness during the cold pressor. Age was significantly related to children's ratings on the self-report measures; younger children reported lower levels of pain intensity and affect than older children. There were no gender differences on the majority of the pain measures, with the exception of facial expressivity, where boys evidenced higher scores than girls. These results indicate that maternal behaviour can have a direct impact on girls' subjective reports of pain and provide evidence of the importance of social learning factors in influencing children's pain experiences. Future research is needed to extend this research to samples of children experiencing chronic pain, as well as explore the possible role of child characteristics (e.g., coping skills) as moderators of maternal interaction effects.

碩士
國立臺灣科技大學
應用外語系
106
As globalization and information technology advance, access to foreign films and TV programs has multiplied. However, the number of studies on subtitle translations has not increased accordingly. House of Cards, an online streaming series by Netflix, has changed not only the fate of the company but also the viewing habits of the audience. This thesis examines two different subtitle translations of the 13 episodes in season one in light of important translation theories and methods. Chapter One introduces the background and motivation, purpose, and outline of the study. Chapter Two is dedicated to the series itself, including its cast, carrier (Netflix), plot, synopsis, and two translators (Deltamac and YYeTs). Chapter Three elaborates on translation methods and theories such as the three requirements of translation, translation universals, domestication and foreignization, skopos theory, equivalence theory, modulation, translation of metaphors, strategies to render ECR, and constraints of film subtitling. In Chapter Four, the researcher analyzes the two versions and selects 21 excerpts to exemplify the appropriate theories and propose her own suggested translations. The strengths and weaknesses of each version are also included in this chapter. Chapter Five concludes with findings.

碩士
國立高雄科技大學
應用英語系
107
This study aims to explore and identify the different translation strategies adopted by YouTubers that are against the norms of subtitle translation in rendering the subtitles of their videos. By analyzing the collected data and comparing currently adopted translation strategies with the norms of subtitle translation, this study aims to understand why some YouTubers tend to adopt certain translation strategies in ways that are shifting from the norms when there are better alternative solutions. This research applies Jan Pedersen’s model of subtitle translation for specific cultural terms and Peter Newmark’s translation procedures to form a strategic analysis framework consisting of seven strategies, i.e., retention, specification, substitution, official equivalent, annotation, tone transference and liberalization of subtitle format. The translation typology proposed by Roman Jakobson is adopted in this research in analyzing the collected data to provide a more comprehensive view of AVT. The norms of subtitle translation in this research are based on a textbook of translation, Netflix style guides, and general knowledge of translational norms. In the process of data analysis, the norms of subtitle translation are used as a standard for the comparison between currently adopted strategies and the norms of subtitle translation. This research finds that use of annotation, retention, tone transference and liberalization of subtitle format are completely against the norms. Among these, annotation is the most frequently adopted translation strategy. Annotation adds additional information and makes target text very lengthy. Strategies conforming to the norms of subtitle translation, like specification, are adopted for different purposes, such as advertising and introducing foreign cultures. To sum up, the present research investigates the differences between the adoption of these aforementioned seven strategies as compared with the norms of subtitle translation, and the findings show that the adoption of specific translation strategies that go against the norm is greatly influenced by what the purpose of the video in question is. In other words, the adoption of a translation strategy that is different from the norms serves specific translation purposes, which confirms the Skopos theory.

碩士
國立高雄第一科技大學
應用英語系口筆譯碩士班
103
Thanks to the advancement of technology and entertainment industry, audio-visual translation has become a popular topic of research in the recent years. Right now in Taiwan, many researchers have already written theses about audio-visual translation, but none of these theses is about slangs used teenagers in the movies, and none of these researchers is done from the perspective of subtitle translation. Usually these slangs used by teenagers are concise and humorous, and thus shows the creativeness in languages and the features of colloquial expression. In order to understand the slangs used by teenager and the features of these slangs, the research has collected the English and Chinese subtitle from 12 teenager comedy that was produced by Hollywood during the period of 2000-2010. The reason teenager comedy is chosen from this decade is that it is the heyday of teenager comedy. Many were produced, and the story and plot is wide-ranged, which includes relationship with lovers, families and friends, thus providing a plentiful corpus of slangs, also, these slangs are the ones that we would frequently in daily life. This research will mainly focus on the domestication or foreignization strategy that was used in the subtitle translation of the 12 teenager comedy, and it wants to find out which strategy was preferred by the translators of subtitle translation and the reason. The research will also focus on Skopos theory, a theory that indicates the purpose dictates how translation works. And the purpose of subtitle translations is to let the audiences understand the plot, especially in comedy, because audiences need to know why the story or dialogue is funny. To let the audience appreciate the humorous effect in comedy, subtitle translators must choose a certain strategy. In the findings of this research, it is observed that most subtitle translators of slangs in teenager comedy preferred domestication strategy. It is due to the fact that some slangs in the ST did not have an equivalent term in target language. If the translator use foreignization strategy, then the translation will not be understandable to the audiences. And concerning the space of the subtitles in the screen, the numbers of words in the subtitle must be limited. The foreignization strategy will produce a wordy translation which is not suitable for subtitle translation. Contrary to that, domestication strategy will produce a concise translation which will convey the humorous effect precisely, and that’s the most important function to subtitle translation.

Some feminist theorists have commented that feminism has succeeded only up to a point - and that point is motherhood. Motherhood, these theorists say, remains one of society's greatest untold stories. This study examines the subjective experiences of one small but important segment of mothers - mothers of a child with autism. The current literature on this topic is mainly focused on the autistic child and only recently has the impact on family members been explored. What relatively little literature exists is focused on "parents", not mothers, thus hiding the gendered nature of care, and is mainly quantitative, anecdotal, and focused on the impact of caring for children with generalized "special needs" - not on children with autism. To date, there has not been a comprehensive gender-sensitive qualitative study about the experiences of mothers caring for a child with autism. The goal of this study is to better understand the personal experiences of these mothers in terms of their interaction and relationship with their support system. Information was gathered through the use of semi-structured interviews with eight mothers. Analysis of their transcripts revealed three themes. Mothers experienced and felt: (1) ignored; (2) abandoned; and (3) like burdened "Supermoms."

Fisheries are often referred to as a common pool resource (CPR) in which exclusion is difficult and resource use involves subtractability. In the estuary of the Patos Lagoon, Southern Brazil, artisanal fisheries management has faced a crisis which brought forth a co-management regime, represented by a Forum, to manage fisheries CPRs. The objectives of the thesis were to analyse the implementation of the Forum arrangement and to recommend ways to strengthen the comanagement process. The analysis centred on the evaluation of the process of decision making for joint use; the role of science and local knowledge in institutional learning; and the congruence between environmental institutions and the conservation of fisheries CPRs. Methods of investigation involved literature review, document analysis, open-ended interviews, closed-ended survey, and participation at the Forum meetings. The Forum represents a move towards a sharing of responsibility and authority on local fisheries management supported by government decentralisation policy. Different stakeholders are now locally involved in the governance of artisanal fisheries in the region. However, the devolution of power to fisher communities is still hampered by the weak involvement of fishers in the Forum and by external influences at government levels. The creation of the Forum has attempted to minimise the problem of fit between institutions and resource conditions. Mismatches still persist, particularly in the definition of access boundaries to resource use, in the design of harvest practices adapted to environmental characteristics, and in the lack of a broader systemic approach to fisheries management. In this context, this study proposes a larger role for fishers knowledge in the comanagement regime, given its potential contribution to the design of sustainable fishing practices for the region. Accomplishments were observed. The Forum has had important outcomes and gained legitimacy. Small-scale fisheries management has become a more transparent process in which management rights have become explicit and openly discussed. The Forum represents a transition in resource management paradigms towards one that is more participatory, in tune with the functions of ecosystems and based on social mechanisms that facilitate the exchange of knowledge and information necessary to build a resilient socialecological system.

The present study analyzed the negative scale of the Attributional Style Questionnaire (ASQ), a popular measure of attributional style and correlate of depression (Peterson et al., 1982), using non-parametric item response modeling. This modeling technique was used to determine if items function differentially for males and females, through differential item functioning (DIF) analysis. Past studies of the ASQ have not used non-parametric item response theory (IRT) modeling in their analysis, nor have they determined whether or not items function differentially for males and females. On a methodological note, nonparametric IRT results were therefore compared with results obtained using classical test theory (CTT) statistics. Results indicated that the negative items of each of the three subscales of the ASQ, locus, stability, and globality were found to function well and discriminate best at the extremes. Further, no gender DIF was detected using either nonparametric IRT or CTT methods. Furthermore, conditional reliability was found to be more informative and powerful than a single reliability coefficient. Findings add to the existing literature supporting the theoretical model of the ASQ by utilizing previously unexplored statistical methodologies.

碩士
國立高雄第一科技大學
應用英語系口筆譯碩士班
104
Culture-bound terms are a significant issue in subtitling translation, since subtitlers need to deal with two languages and cultures concurrently within the temporal and spatial restrictions in subtitles. In this sense, rendering strategies for culture-bound terms are usually deployed to produce translations that help the target audience to understand foreign culture-bound terms. Most of the current strategy models of culture-bound terms are developed from Western language systems, which might not be viable in other language systems. Consequently, this thesis focuses on investigating whether these strategy models of culture-bound terms could be applied in Chinese subtitles. With regard to the general subtitling strategies proposed by Gottlieb (1992), this thesis compares the types of strategies devised by Nedergaard-Larsen (1993), Gottlieb (2009), Díaz Cintas and Remael (2007), and Pedersen (2011), and later adopts Pedersen’s model as its analytical framework, since it is currently the most complete model. His model includes the strategies of retention, specification, direct translation, generalization, substitution, omission and official equivalent. These strategies are used to examine examples from fifteen English-language films. The results show that the strategy model proposed by Pedersen (2011) may not be sufficiently applied in Chinese subtitles. First of all, the use of homophones is a characteristic of the translation of culture-bound terms in Chinese subtitles, which is not included in his model. Next, the use of direct translation and situational substitution are not found in the collected data, indicating that these strategies might not be used frequently in Chinese subtitles. Moreover, problems such as credibility gaps and negative effects are caused in some cases of Chinese subtitle translation. Fortunately, such problems can be solved by applying the strategies of specification, generalization, substitution, and official equivalent. The data analysis shows that generalization is mostly used to solve these problems. Consequently, this thesis proposes a new model based on Pedersen’s strategies and modified with the characteristics of culture-bound terms in Chinese subtitles. In the new model, the strategies of direct translation and situational substitution are eliminated but the strategy of homophone is added. This model sheds light on subtitling of culture-bound terms from a Chinese linguistic perspective in Translation studies.

碩士
國立高雄第一科技大學
應用英語系口筆譯碩士班
103
Abstract With the development of economic globalization, intercultural communication is in a rapid growth, among which films constitute an important channel. As the foreign films pour into Taiwan, Taiwanese films also begin to enter the international market. As a medium of information transmission, film subtitle translation plays a more and more important role. A high-quality subtitle translation can attract more audience and better convey cultures. As a new field of translation, film subtitle translation has attracted a great deal of interest among researchers and become a current focus for research. In Europe and America there has been an upsurge in studies of multimedia translation, translation strategies and quality control of subtitle translation. However, domestic studies on subtitle translation haven''t caught sufficient attention and have been lagging behind, which contributes negatively to the quality of Taiwanese film subtitle translation. This research uses five translation strategies (direct translation, paraphrase, cultural substitution, adaptation, and omission) to examine slang found in four Taiwanese films. According to the findings, the translators of these films preferred to use mostly the paraphrasing strategy in translating Tai-yu slang. The frequency of using this strategy is about 60 percent. This strategy makes it possible to express the translated text in a plainer and simpler way than the source text so that the original implication between the lines can be easily revealed.

碩士
國立雲林科技大學
應用外語系
102
The purpose of this study is to examine the phenomenon of simplification in the subtitle translation of the movie: The Diary of a Wimpy Kid: Rodrick Rules. In order to present an objective analysis, the researcher adopted Gottlieb’s (1992) subtitling strategies, including expansion, paraphrase, transfer, imitation, transcription, dislocation, condensation, decimation, deletion and resignation. This study examined the traditional and simplified Chinese subtitles. To this end, this research focused on the use of the three simplification strategies (including condensation, decimation and deletion). The findings revealed that the most frequently deleted utterances of all were pronouns followed by pet phrases. The study also found that the condensation and decimation strategies were used more frequently for the traditional Chinese subtitles than for the simplified Chinese subtitles, and that the difference in the frequencies between the two subtitling versions was significant. In addition, two of the simplified Chinese subtitles were found unable to be classified into any of Gottlieb’s (1992) strategies because the words were simply transcribed into the Chinese subtitles. This direct transfer from the source text to the target one was named the strategy of “carry-over.” Finally, the words and phrases which are omitted or simplified in both of the Chinese subtitles might be used in comparison with other subtitle corpuses for translation analysis.