This study represents a sociolinguistic contribution to the analysis of Past Participle Agreement (PPA) in [avoir+PP] constructions. In spoken French, this agreement is marked only for gender, and is subject to much variation in the production of speakers of French. The objectives of this thesis are twofold: firstly, to observe the variable in the context of production, by a quantitative study of several corpora. Variation in the production of PPA is tested against linguistic factors, such as the position of the participle in the verbal group where it appears. Social and stylistic factors are also considered, notably the socioprofessional category of the speaker. These analyses reveal that agreement is conditioned by a complex interaction of these factors. The second objective is the design of a Matched Guise Test, based on scripted conversations. This test was carried out with a view to measuring the extent to which the presence or absence of marked PPA, often considered a typical result of normative pressures, affects the stereotypical social representation of a speaker. Differences in informants' judgements were modest, and two issues are discussed with regard to this: firstly, the validity of the methodology, and secondly, the evaluation of PPA as a sociolinguistic marker. Both parts of this thesis reflect the methodological issues pertaining to the investigation of a rare variable.

Submitted by Maike Costa (maiksebas@gmail.com) on 2016-07-22T11:31:47Z No. of bitstreams: 1 arquivo total.pdf: 2482366 bytes, checksum: 61b9b71ecd2bb111b320025bf6adab3c (MD5) / Made available in DSpace on 2016-07-22T11:31:47Z (GMT). No. of bitstreams: 1 arquivo total.pdf: 2482366 bytes, checksum: 61b9b71ecd2bb111b320025bf6adab3c (MD5) Previous issue date: 2012-12-20 / This thesis aims at analyzing the stylistic aspects of Brazilian sign language in use in the cities of Recife and João Pessoa, considering its variable feature in a sociolinguistic context. We assume that the sociolinguistic variation is a means through which the social is incorporated into the language and therefore the nature of social meaning is valued. Thus, variation becomes meaningful. In methodological procedures we consider that it is a documentary research, which made use of a database project entitled Preliminary Studies for the National Inventory of Linguistic Diversity - The Brazilian Sign Language in the Northeast: the variants of João Pessoa and Recife. The analysis and discussion of the results were developed with the support of the software ELAN, a relevant tool for the transcription and analysis of data. At first, sociointeractionist analysis with 96 (ninety-six) users of LIBRAS was done, 54 (fifty four) living in Joao Pessoa and 42 (forty-two) in Recife. As a second step, from the 200 recorded signals in the Swadesh list, we chose five of them that showed a large lexical variation for analysis with ELAN: WHITE, HUSBAND, WIFE, CHILD and PERSON. The third moment was devoted to the study of the dialect in the community, that is, how these signals now are used in the context of informal interaction and communication. As research results we can consider: where and how the language circulates among its users, as well as its purpose, relevance, age and place of acquisition / learning, the interpreter’s needs in the contexts of social interaction, the (in) existence of instructional material, cultural groups and cultural material in LIBRAS, its fluency in everyday life and language difficulties with its user. In the analysis focused on style, we analyzed aspects that are particular of LIBRAS, such as configuration parameters of the hand, point of articulation, movement, orientation and facial and body expressions, considering the dialectal variation between towns (João Pessoa and Recife), as well as the dialect in the community in practice, in the context of use by subjects. Finally, we described and analyzed the variants, from an informal to a formal condition of annunciation and thus we valued the unique style that LIBRAS evidences, which assures that even as a signaled language it is full of indexical prints and value in the Sociolinguistics field. / A presente tese teve como objetivo analisar os aspectos de estilo da Língua Brasileira de Sinais em uso nos municípios de João Pessoa e de Recife, considerando seu aspecto variacionista em um contexto sociolinguístico. Partimos do pressuposto que a variação sociolinguística é um meio através do qual o social é incorporado à linguagem e, por isso, a natureza do significado social é valorizada. Assim, a variação passa a ser repleta de significado. Nos procedimentos metodológicos consideramos que se trata de uma pesquisa documental, a qual fez uso de um banco de dados inerente ao projeto intitulado Estudos Preliminares para o Inventário Nacional da Diversidade Linguística – A Língua Brasileira de Sinais no Nordeste: as variantes de João Pessoa e Recife. A análise e discussão dos resultados foi desenvolvida com o apoio do programa ELAN, instrumento relevante para a transcrição e análise dos dados. Em um primeiro momento, foi feita uma análise sociointeracionista frente aos dados coletados de 96 (noventa e seis) usuários da LIBRAS, sendo 54 (cinquenta e quatro) residentes em João Pessoa e 42 (quarenta e dois) em Recife. Em um segundo momento, dos 200 sinais registrados na Lista de Swadesh, elegemos cinco deles que evidenciaram uma grande variação lexical: BRANCO, ESPOSO, ESPOSA, CRIANÇA e PESSOA para análise por meio do programa ELAN. O terceiro momento, por sua vez, voltou-se ao estudo do dialeto na comunidade, isto é, como esses sinais, agora, são usados no contexto informal de interação e comunicação. Enquanto resultados da pesquisa, podemos considerar: abordamos onde e como a língua circula em meio a seus usuários, bem como sua finalidade, relevância, idade e locais de aquisição/aprendizagem da LIBRAS, necessidade ou não de intérprete nos contextos sociais de interação, a (in) existência de material instrucional, grupos culturais e material cultural em LIBRAS, fluência desta no dia a dia e dificuldades como usuário desta língua. Na análise voltada ao estilo, analisamos aspectos que são particulares da LIBRAS, tais como os parâmetros de configuração de mão, ponto de articulação, movimento, orientação e expressões facial e corporal, considerando a variação inter e intra dialetal entre dois municípios (João Pessoa e Recife), assim como o dialeto na comunidade, na prática, no contexto de uso pelos sujeitos. Por fim, descrevemos e analisamos as variantes, desde uma condição formal à informal de anunciação e, assim, valorizamos o estilo próprio que a LIBRAS evidencia, o que nos garante que, mesmo sendo uma língua sinalizada, ela é repleta de marcas indexicais e valores no campo da Sociolinguística.

This thesis investigates phonetic and phonological variation in the bilingual repertoire of adolescent Welsh-English bilinguals living in North Wales. It contributes to linguistic research by, firstly, providing an account of language variation in an under-studied area (N. Wales) and context (regional minority language bilingualism) and, secondly, by examining cross-linguistic variation, and the constraints on this variation, in bilingual speech. The two variables under discussion differ in how they are realised in the two languages: /l/ is thought to be heavily velarised in both languages as a result of long-term contact and phonological convergence. Variation in the production of /r/ and realisation of coda /r/ has hitherto been reported as language-specific, though frequent transfer is said to occur from Welsh to English in predominantly Welsh-speaking areas (e.g. Penhallurick 2004: 110; Wells 1982: 390).The first aim of the study is therefore to quantify claims of phonological convergence and transfer in the speech of Welsh-English bilinguals by using a variationist sociolinguistics methodology (e.g. Labov 1966), which also considers the influence of linguistic and extra-linguistic factors on variation. Particular attention is paid to differences between a majority Welsh-speaking town and a town where English is the main language. A further distinction is made between those from Welsh-speaking homes and those from English-speaking homes who have acquired Welsh through immersion education. The second aim is to make empirically-informed theoretical claims about the nature of phonological convergence and transfer, and conceptualise cross-linguistic interaction in the speech of Welsh-English bilinguals in light of existing frameworks. Data (sociolinguistic interviews and wordlists) were collected in Welsh and English from 32 Welsh-English bilinguals aged 16-18. The sample was equally stratified in terms of speaker sex, home language, and area. The two towns compared in the study are Caernarfon (N.W. Wales, where c.88% of the population speak Welsh) and Mold (N.E. Wales, where c. 20% Welsh of the population speak Welsh). The results indicate that English [ɫ] tends to be lighter than Welsh [ɫ] in word-initial onset position for females, and in word-medial intervocalic position for both males and females. The data also show linguistic influences on the realisation of [ɫ] in both languages, and differences between males and females. The realisation of coda /r/ and production of [r] and [ɾ] in English are confined to the speech of those from Welsh-speaking homes in Caernarfon. In Welsh, use of [ɹ] is widespread and is constrained by a more complex interaction between area, home language, and sex. On the basis of these findings, I conclude that features which have undergone phonological convergence due to long-term language contact may be subject to language-specific constraints when implemented phonetically. In terms of transfer, I argue for a ternary distinction between interference, transfer, and transfer which is constrained by linguistic and/or extra-linguistic factors (cf. Grosjean 2012). Finally, I suggest that Mufwene’s (2001) notion of the ‘feature pool’ is the most succinct way of conceptualising Welsh-English transfer and differentiate between more focussed accents of English and a less-focussed variety of North Wales Welsh.