On the subject of what affect voters¡¦ vote choice, political scientists for a long time emphasize three answers: party identification, candidate orientation, and issue orientation. About issue orientation, Rational Choice Theory assumes that human are rational pursuing maximizing self- interests. When voters are making their vote decisions, they would observe the issues positions of competing parties or candidates, comparing with their own positions, and then vote the party or the candidate who can represent their own positions best. Spatial Theory, from Rational Choice Theory, takes those abstract issue positions into some issue space. Both the issue positions of voters and parties could be presented by some points in the space, and the length and direction between the points can represent the differences between issue positions. There are several different models in Spatial Theory, and different models advocate different ways about how voters use the points in issue space to form their evaluations to competing parties or candidates. In this paper, we take the viewpoints of Spatial Theory to research the issue voting behavior of Taiwanese voters. First, we try to know the spatial distribution of voters¡¦ issue positions. Then we inspect the association between voters¡¦ social back- ground elements and issue position. Finally, we test three models of Spatial Theory, proximity model, directional model, and RM mixed model, to know how Taiwanese voters use issue positions to form their party-evaluation.

Submitted by Cristiane Chim (cristiane.chim@ucpel.edu.br) on 2018-07-16T12:23:03Z No. of bitstreams: 1 Camila Motta Avila.pdf: 5908666 bytes, checksum: fec1aaf83edeb30b980911d427091061 (MD5) / Made available in DSpace on 2018-07-16T12:23:03Z (GMT). No. of bitstreams: 1 Camila Motta Avila.pdf: 5908666 bytes, checksum: fec1aaf83edeb30b980911d427091061 (MD5) Previous issue date: 2017-12-21 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - CAPES# / #2075167498588264571# / #600 / This Master’s Dissertation had as its main goal analyzing and formalizing how Brazilian learners of English language (L2) and Americans identify and discriminate the voicing patterns for plosive consonants under artificial manipulation in English monossilabic words. Americans normally follow the acoustic cue Voice Onset Time (VOT) to differentiate voiced plosives from voiceless ones. The VOT is measured in milisseconds and can be classified into three different categories (LISKER & ABRAMSON, 1964): negative VOT (for /b/, /d/ and /g/ in Portuguese); short-lag VOT (for /p/, /t/, /k/ in Portuguese and /b/, /d/, /g/ in English) and long-lag or positive VOT (for [ph], [th], [kh] in English). Previous studies (ALVES, BARATZ e MOTTA, 2012; SCHWARTZHAUPT, ALVES &FONTES, 2015; ALVES & MOTTA, 2014; MOTTA, 2014; ALVES & ZIMMER, 2015) demonstrated that the manipulation of VOT can result into different types of perceptual behaviour, depending on the group analyzed: Americans tend to follow this cue to attribute voicing patterns to consonants, while Brazilians tend to follow the voicing patters stablished for their L1. In order to analyze the perception in L2, this Dissertation was based on Speech Learning Model (SLM) – (FLEGE, 1995) and Perceptual Assimilation Model – L2 (PAM-L2) – (BEST e TYLER, 2007) studies. To investigate how perception was processed in Americans, explanations in Kent and Read (2015), Gussenhoven and Jacobs (2004), Matzenauer (2015) and Boersma, Escudero and Hayes (2003) were found. The method consisted of two perceptual tests: (a) Identification Test, (b) Discrimination Test. Both tests were designed with artificially manipulated stimuli. The types (pee, pit, tick, tip, kit, kill) were CVC (where V was a high vowel /ɪ/ or /i/). Each one of them was multiply manipulated, generating five tokens from one original type. This manipulation was gradually performed, in 25% layers, aiming to achieve the zero artificial VOT pattern. The Identification Test was designed with 60 target words and 12 distractors, resulting in 72 tokens per participant. The Discrimination Test was designed with 36 target trials and 9 catch-trials, in a total of 45 trials per participant. To summarize, results show that Americans can be considered to be more perceptually sensitive to the multiple VOT manipulation than Brazilians, who normally tend to follow their L1 way of attributing voicing patterns to plosives, regardless their proficiency level in English (basic, intermediate or advanced). In addition, different phonetic tresholds were found in each analyzed group and in each place of articulation observed (bilabial, alveolar, velar). Finally, this study tries to offer a formalization for perceptual grammars in each group, based on Bi-directional Phonology OT Model (BiPhon). It is expected that this study, its detailed data description and theoretical observations can contribute to and makefurther academic studies possible. / Esta Dissertação de Mestrado teve como principal objetivo analisar e formalizar de que forma brasileiros aprendizes de inglês (L2) e americanos identificam e discriminam o vozeamento das consoantes plosivas iniciais da língua inglesa sob múltiplas manipulações em palavras monossilábicas. Na língua inglesa, para a diferenciação de vozeamento, tem-se como principal pista acústica a aspiração, que pode ser medida a partir dos valores de Voice Onset Time (VOT). O VOT é medido em milissegundos e pode ser classificado em 3 diferentes padrões (LISKER & ABRAMSON, 1964): VOT NEGATIVO (que se realiza foneticamente em português em /b/, /d/ e /g/); VOT ZERO, (que se realiza como /p/, /t/ e /k/ no português e /b/, /d/ e /g/ no inglês) e o VOT POSITIVO (encontrado em [ph], [th] e [kh] no inglês). Estudos anteriores (ALVES, BARATZ e MOTTA, 2012; SCHWARTZHAUPT, ALVES &FONTES, 2015; ALVES & MOTTA, 2014; MOTTA, 2014; ALVES & ZIMMER, 2015) demonstraram que a manipulação da pista acústica VOT resulta em comportamentos perceptuais diferenciados de acordo com a natureza do participante: americanos tendem a seguir a pista acústica em questão ao atribuir ou não vozeamento para as consoantes plosivas, ao passo que os brasileiros continuam a seguir os padrões de vozeamento previstos para a sua L1. Para dar conta da percepção em inglês como L2, este trabalho fundamentou-se, basicamente, nos modelos perceptuais Speech Learning Model (SLM) – (FLEGE, 1995) e Perceptual Assimilation Model – L2 (PAM-L2) – (BEST e TYLER, 2007). Para analisar a percepção dos participantes americanos, buscaram-se explicações em Kent e Read (2015), Gussenhoven e Jacobs (2004), Matzenauer (2015) e Boersma, Escudero e Hayes (2003). O método aplicado neste estudo consistiu em dois testes de percepção: (a) Teste de Identificação, (b) Teste de Discriminação. Ambos foram montados com estímulos de palavras monossilábicas da língua inglesa (CVC), sendo V /ɪ/ ou /i/. Cada type (pee, pit, tick, tip, kit, kill) passou por manipulação múltipla da pista acústica VOT, gerando 5 tokens cada um. Tal manipulação se deu de forma gradual em camadas de 25%, objetivando-se alcançar o padrão de VOT zero artificial. O design do Teste de Identificação contou com 60 palavras-alvo e 12 distratoras, somando um total de 72 tokens por participante. O design do Teste de Discriminação, que teve por objetivo contrastar diferentes camadas de VOT entre si, contou com 36 tríades do tipo AxB e 9 catch-trials, somando 45 tríades por participante. Sucintamente, os resultados apresentados demonstraram que os participantes americanos são mais sensíveis ao corte da pista acústica VOT do que os participantes brasileiros, independentemente do nível de proficiência em língua inglesa (básico, intermediário e avançado). Além disso, limiares fonéticos distintos foram encontrados em cada um dos grupos analisados e em cada ponto de articulação observado (bilabial, alveolar, velar). Finalmente, propôs-se uma formalização da percepção de ambos os grupos para cada consoante analisada baseando-se no Modelo de OT Bidirecional (BiPhon), proposto por Boersma (2009) e Boersma e Hamman (2011). Espera-se, com este estudo, que a descrição detalhada dos dados e a reflexão teórica possam contribuir à Academia e possibilitar estudos futuros.