碩士 / 國立新竹教育大學 / 臺灣語言與語文教育研究所 / 92 / As stated in Chapter 1, the purpose of the thesis is to give an explicit definition of serial verb construction (SVC for short) in Taiwanese Southern Min (TSM for short) with respect to its scope, constituents, types, grammatical function and syntactic representation. In particular, we will provide an in-depth analysis of related issues.　On the strength of historical records Chapter 2 highlights the development of SVCs as revealed in the texts from Yin-Shang period through Qin-Han period to Tang-Song period in an attempt to uncover the traits of SVC in comparison with its counterpart in modern Chinese. We also give a literature review of previous works published at home or abroad. Chapter 3 fleshes out the characteristics of SVC in the following terms. SVC is a compound sentence denoting a special coordinate relationship. The construction is made up of at least two predicates sharing the subject and features neither conjunction or complementizer nor syntactic relationship such as modifier-modified, verb-complement, and verb-subject/object in the pivotal construction. This chapter also teases out the grammatical features of SVC, the diagnostic principles, construction types and sentence patterns, and makes a comparison of SVC and other special constructions denoting simultaneous actions. Chapter 4 gives a fine-grained analysis of sentential constituents that function as subject, first predicate, second predicate and object. Following a look into the syntactic aspect of SVC Chapter 5 arrives at a global and thorough understanding of SVC in TSM by delving into its grammatical function, semantic relationship and traits of representation. Chapter 5 enlists the commonality and difference between SVC and other types of construction so that SVC becomes more clearly delineated and its status is therefore more firmly established. Chapter 7 as an extension of the above endeavor examines the SVC in TSM proverbs from which interesting phenomena appear such as the syntagmatic relationship of verb, adjective, object and complement, different sentence patterns and various kinds of semantic relationship. Chapter 8 addresses overwhelming issues that crop up in our analysis such as deictic verbs lai5 and khi3, the definition of preposition, progressive aspect and complicated SVC. Chapter 9 summarizes the discussion of this thesis and briefly touches on the contrastive aspects of SVCs as shown in TSM and Mandarin. Furthermore, we advance some residual unsettled questions such as the thorny problems of giving a precise and comprehensive definition of preposition to pin down the dual role of words in SVCs, the juncture and tone sandhi in a sequence of predicates, the durations of predicates in SVC and other constructions, frequency of predicates and kinds of SVCs and the frequency of semantic types and dialectal divergence of SVC. All of these questions form important points of departure in our future exploration.

碩士 / 國立高雄師範大學 / 英語學系 / 97 / This thesis aims to explore the Taiwanese secret languages, Banana colloquial speeches. The data are first gathered by fieldwork and then compared with other relevant secret languages to draw the characteristics of Banana colloquial speeches. Finally, their formations are analyzed under OT framework. The fieldwork surveys are conducted in Neimen of Kaohsiung county as well as Hsingang of Chiayi county, in which data of three informants are collected. Their data are designated as Kaohsiung Banana colloquial speech and Chiayi Banana colloquial speech. They are then compared with other previously recorded La-Mi type secret languages, including Cantonese La-Mi (Chao, 1931), Wun-tsyu-pe-a-we (Li, 1985), and Tainan Banana colloquial speech (Yang, 1996). The similarity shared by them is that they all involve reduplication which makes one base syllable into two syllables and the first one is fixed with [l] in the onset position while the second one fixed with [i] in its nucleus. The differences among them exhibit in the following aspects: nasality spreading, coda neutralization, onset dissimilation, rime dissimilation, syllabic nasal, and tone sandhi. However, all named as Banana colloquial speeches, these speeches have more common features, especially in their dealing with onset dissimilation, rime dissimilation, and syllabic nasal when compared to the other secret languages. OT analysis provides the following accounts for Banana colloquial speeches’ formation. First, the reduplication process involved is realized by self-compounding which can be derived through interaction of constraints like ALLITERATE, RHYME, and MAX-BD. Second, the fixed segments in the two resultant syllables fit into the two types of fixed segmentism proposed by Alderete at al. (1999). [l], in the first onset, and [i], in the second nucleus, belong to the morphological type and can be accounted for by the constraint ranking of REALIZE MORPH dominating MAX-BD. In contrast, [t]/[n] in the second codas surface as phonological type fixed segments, and are derived through the constraint ranking of markedness constraints with other faithfulness constraints. Third, the differences among these secret languages, such as their treatments of dissimilation, can be explained by the different rankings of the same set of constraints.

碩士 / 國立新竹教育大學 / 臺灣語言與語文教育研究所 / 94 / The present study is aimed at investigating how the subjects process changed tonal range in Taiwanese. Two event-related potential（ERP）experiments are designed to delineate the unit in which the tonal range could be changed. In experiment 1, we investigated the minimal range within which the subjects can discriminate among the three level tones. The results show that the minimal tonal range in Taiwanese is at least four semitones. The boundary between the mid tone and the high tone is located at one semitone over the mid tone. The lower boundary of the mid tone is located at two semitones below the mid tone. However, in the range of low level pitch, the mid tone and low tone become indistinct. Nonetheless, the falling pattern is an important feature of the low tone in Taiwanese, so that the low level tone used in our experiment is not distinguishable from the mid tone in the range beyond the lower boundary of the mid tone. Gradual differences in perception are shown in the ERP data. When the pitch played to the subjects is beyond the upper and lower boundary of the tone in question, the more deviant the easier in processing, inducing smaller N400; when the pitch played to the subjects is within the upper and lower boundaries of the tone in question, the more deviant the more difficult in processing, inducing larger N400. The largest N400 amplitude is induced in the frontal-parietal area. In experiment 2, we tried to find out what kind of unit, with changed tonal range, is acceptable to the subjects. Four types are manipulated. Each type contains two conditions: sentences with and without changed tonal range. ERP data were recorded from the starting point of changed tonal range. Compared to the sentences without changed tonal range, the ones with changed tonal range in type 1, 2, and 4 induce N400. It seems to suggest that the 3 type, i.e. the changing of the tonal range in the whole tone sandhi group induces less difficulties in processing. The largest N400 amplitude is induced in the frontal-parietal area. Key Words: tone, tonal range, ERP, N400

碩士 / 國立中央大學 / 客家語文研究所 / 101 / The Hakka dialect in Sichuan is the product of the immigration policy during Ming and Ching Dynasty, so far, it has become one of the influential Chinese dialects in Sichuan. This thesis emphasizes on the hakka dialect of Tashui Zhen(踏水鎮),a town in Sichuan Jianyang(簡陽).The materials of Tashui hakka dialect are investigated and written, Tashui hakka dialect ,which came from Guangdong Wuhua(五華), we discuss the phonological and lexical characteristics of Tashui dialect ,that shows the differences between the origin country Wuhua and the new settlement Tashui. The Mandarinization of Tashui dialect reveals a unique point that attracts our attention, we regard it as the product of language contact because the intense interaction has shown us the compromise with the southwest Mandarin. This thesis is divided into seven chapters, it contains phonology and lexicon discussion, but we mainly focus on the former, the latter is auxiliary. The first chapter is the introduction which explains the motivation ,purpose, research method, procedure, bibliography of related articles, as well as brief situation of Sichuan history and geography, where hakka dialect spread in Sichuan, and simple information of utterance speakers. The second chapter represents the phonetic system of Tashui dialect, the initials, finals, tones and the special tone sandhi. The historical phonology is shown in Chapter three, we compare the Tashui dialect with middle Chinese phonetic system, after comparing, all the phonological characteristics contribute to the fourth chapter.we found out that some features remains, such as sang tone (上聲) in ancient Qu (去聲) with voiced initial. However, it also influenced by southwest Mandarin (西南官話), such as in finals (陰聲韻)with yang finals(陽聲韻)and Velar nasal consonant “-“ appears in ing (影母). The comparison between Wuhua and Tashui dialect is discussed in chapter five, including phonology, lexicons, even the morphology, generally speaking, the massive change alters beyond recognition. The chapter six talks about the Tashui lexicons, hakka dialectal feature words are still used, but the obvious Mandarinization phenomena show a big impact on Sichuan hakka dialects. Briefly speaking, there are five lexical features in Tashui dialect：(1) the impact of language contact(2) various lexicon statement and detail categories (3)cultural change (4)environmental constraint (5) reduplication word formation. The Chapter seven is the conclusion, to sum up the phonology and lexical features of Tashui dialect, Finally, an examination of the shortcomings of this study and possible future advices are provided to make further development of Sichuan hakka dialects.

Research on prosodic structure, the underlying structure organising the prosodic grouping of spoken utterances, has shown that it consists of hierarchically organised prosodic constituents. The present thesis explores the nature of this constituency, in particular the question of whether prosodic structure is comprised of a given set of qualitatively distinct domains, or of a set of domains of the same type varying only gradiently in "strength", or a possible mixture of both types of relations across prosodic levels. This question is addressed by testing how prosodic constituency (mirrored on boundary strength manipulations) is signalled acoustically via pre- and post-boundary durations, intonation contours, and two sandhi processes, namely vowel hiatus resolution and post-nasal stop voicing in Modern Greek. Results show that the phonetic signalling of boundary strength provides support for a mixture of both differences of type and strength across prosodic levels, with some levels only differing in terms of their strength. Pre-boundary durations and resolution of vowel hiatus are gradiently affected by boundary strength with shorter to longer durations from lower to higher domains, and less instances of vowel deletion higher in the hierarchy. Post-nasal stop voicing is qualitatively affected by boundary strength with almost all voicing instances occurring in the lowest constituent of the structure in the way a qualitative view of prosodic constituency would predict, and in line with research on prosodic phonology. Finally, both the alignment and scaling of intonation contours at the edges of domains is found to distinguish qualitatively the lowest domain from the higher ones. All higher phrasal domains align with respect to the boundary and their peak scaling varies consistently gradiently across speakers. When combining those two findings, support is provided for the existence of differences of strength and type within the same process. Taken together the results from these four phenomena support the postulation of an underlying prosodic structure with a limited number of qualitatively distinct domains, within which at the same time some type of recursivity or structured variability must be allowed for. It is shown that there are structural properties of speech, like the length of the utterance, influencing the organisation of utterances in a principled gradient manner, supporting the existence of differences of strength within domain types. These findings bear significance for theories of prosodic structure that have assumed either the view of solely qualitative differences, or sole boundary strength differences, as well as for future proposals on prosodic constituency. Finally, the use of Modern Greek in this thesis adds to the existing literature on a language that has been extensively used by researchers working in views supporting the existence of qualitative distinctions of type across prosodic domains, and provides the first in depth experimental analysis of post-nasal stop voicing.

This study identifies two syntactically distinguishable types of Serial Verb Constructions (SVCs) in Mandarin Chinese (MC) and Jinjiang Southern Min (JSM), corresponding to the nuclear and core distinction made in Role and Reference Grammar (Foley and Van Valin 1984, Foley and Olson 1985, Van Valin and LaPolla 1997). This distinction is also made on the basis of a general consensus of the cross-linguistic classifications of the processes of monoclausal multi-verb construction formation (e.g., Butt 1993, 1997, Baker and Harvey 2010): namely, predicate fusion and argument fusion. In this study, I propose two sets of diagnostics to establish the distinction; these go beyond the range covered in previous studies (e.g., Olson 1981, Foley and Olson 1985, Crowley 2002, Chang 2007). In the first set of diagnostics in this study, seven inter-clausal diagnostics are considered as the threshold where the behaviours of bi-clausal structures and SVCs split. These diagnostics include independent negation, passivisation of the object of V2, independent modification by temporal adverbial, independent marking of viewpoint aspect, independent modification by manner adverbial, prosodic structure and the Coordinate Structure Constraint (Ross 1967) that is employed in a more restricted manner. In the second set of diagnostics, four intra-clausal diagnostics are adopted to make the distinction between nuclear and core SVCs, which include passivisation of O1, insertion of intervening material, coordination within the SVC, and obligatory topicalisation of undergoer argument. Of particular interest is the possibility that the same string of verbs may occur in superficially similar, but structurally different, SVCs: for example, the Cause-Effect SVC and the Excessive SVC. The diagnostics employed in this study are proposed as a novel method to establish the distinction between the SVCs and the bi-clausal structures, and more importantly, between core and nuclear types of SVC. Contributing to the originality of the new method of diagnosing the status of the SVCs proposed in this study, I add five novel diagnostics, such as passivisation of the object of V2, independent marking of viewpoint aspect, tone sandhi between adjacent verbs, coordination within the SVC and obligatory topicalisation of the undergoer argument, in addition to those that have been employed in the literature. I restrict myself to data of MC and JSM in discussing the rationale of the diagnostics. However, this novel method of identifying SVCs is expected to be cross-linguistically applicable with consistent results, while at the same time allowing for the possibility of cross-linguistic differences in the semantic sub-types of SVCs identified in each language.

碩士 / 國立中央大學 / 客家語文研究所 / 107 / The dialect point studied in this paper is located in Guangxi, Xingbin District, Laibin City, Xiaopingyang Town Hakka dialect. Laibin City is located at central Guangxi, Laibin City is located in central Guangxi. And it is also bounded by Guilin to the north, the south, the east, and the west. It is to the north of Liuzhou City, Guilin City, and Hechi City. It is to the east of Quzhou City, Guilin City,and Guigang City. It is to the west of Hechi City and Nanning City. It is to the south of Guigang City and Nanning City.Guangxi is a treasure land with rich language resources. Under the contact and integration of many dialects, it forms the unique characteristics of Hakka dialect in this area. This study takes the Hakka dialect of Xiaopingyang Town in Laibin city as the research object. The paper takes actual survey method to study voice system and voice evolvement phenomenon of Hakka in Xiaopingyang Town and understand the historical phonological features and experienced phonological changes of Xiaopingyang Hakka dialect. With the further exploration of vocabulary, observe the vocabulary features of the Xiaopingyang Hakka dialect, and compare it with the Longhu, Shapu, and Gonghui Hakka dialects in Guangxi, and sort out the differences between vocabulary. There are six chapters in this study. The first chapter is an introduction that consists of the motive and purpose,methodology,literature review.The second chapter represents the phonetic system of Xiaopingyang Hakka dialect that was obtained from the field reseach, including the initials, finals, tones, feature of tonal sandhi, literary and colloquial reading.In the third chapter,a comparison is made between Xiaopingyang Hakka dialect and the ancient Chinese phonology to discuss the sound evolvement and development of Xiaopingyang Hakka dialect.The fourth chapter is the special features of Xiaopingyang Hakka. It is also a review of first three chapters. It reviews initials, finals and tones in order to discuss the Xiaopingyang Hakka dialect situation and change mechanism. The fifth chapter is the discussion of Xiaopingyang Hakka vocabularies. Throught arrangment ,we can classify five major categories of vocabulary, including type A (effected by Mandarin more), type B (the same in Guangxi), type C (unique), type BA(both type B and type A) , type AC(both type A and type C).The last chapter is the conclusion of the research results.It also reviews the shortage of this research and proposes the suggestions and directions for future Xiaopingyang Hakka research.

This study describes the acquisition of phonology by three children learning Cantonese as their native language. Wai, the primary subject in the study, was observed longitudinally for a period of a year. Wing and Ching, the other two subjects, were observed cross-sectionally for purpose of comparison. Unlike other studies of the acquisition of Chinese phonology, this study proposes an explicit set of analytic procedures and criteria for the phonological analysis of children acquiring Cantonese. The analysis uses Ingram (1981) as its starting point. Altogether, I present four kinds of phonological analyses. They are (1) phonetic analysis, (2) analysis of reduplication, (3) substitution analysis, and (4) phonological process analysis. In the phonetic analysis, I look at two aspects of the children's phonetic ability. These are (1) the total number of segmental sounds and the articulation scores, which are the gross quantitative measures of the phonetic ability of the children's speech, and (2) the analysis of individual segments. Based on these results, I set up an inventory of the early sounds acquired in Cantonese. It is felt that such an inventory can be compared to those of other Cantonese subjects and to those of children learning other languages. In the analysis of reduplication, I distinguish two types of reduplicated forms in the subjects' speech. Type A forms are limited to 'baby talk'. The equivalent of this type of reduplicated form in Cantonese is the baby talk 'baabaa' for sheep, or 'kaka' for car in English. Type B forms are cases where the child reduplicated the adult model. In studying the young child's reduplication in Cantonese, it is necessary to look for both types of reduplication to separate the child who actually reduplicates from the one who is simply repeating reduplicated models. The results of the analysis of Type A reduplicated forms are consistent with the findings from English (e.g. Fee and Ingram, 1982) which show that reduplication occurs early and then decreases greatly. It is also found that the greatest number of the Type A forms are nouns. The, second greatest number are verbs. Adjectives/adverbs are the least frequent. The substitution analysis compares the phonology of. the child's words to their adult models to determine the matches and mismatches between them. In order to quantify this, I adopt two measures to calculate the extent to which matches occur. They are the proportion of matches and the proportion of data. The proportion of matches is the number of matches over the total number of adult sounds attempted by the child. The proportion of data is the number of sounds attempted over the total number of possible sounds in Cantonese. Based on these measures, the substitution of sounds of each subject is examined and compared with the others. Moreover, I also establish a hierarchy of difficulty for the individual segments for Wai. As far as the tone substitutions are concerned, the results justify the general finding that the mastery of tones occurs well in advance of the mastery of segments. All the subjects studied made few errors. In Cantonese, it has been proposed that a tone sandhi rule is commonly observed (Chao, 1947). The rule states that when an upper even tone (tone value 53:) is followed by another upper even tone (tone value 53:) or a high entering tone (tone value 5:), the first syllable will become the tone value 55:. The status of this Cantonese sandhi rule, however, has been challenged in J.K.Tse (1978). He looked at the acquisition of tone development of his son for 30 months, and found no instances of the tone value 53. From these data, he concluded that there is no systematic evidence to support the Cantonese sandhi rule described above. In the speech data of the subjects of this study, all the upper even tones produced by the subjects belong to the tone value 55:. There is no incidence of 53: variant. Thus my results support those of J.K. Tse. Bian-yin (changed tone) is also examined. In Cantonese, it is commonly agreed that there are two products of Bian-yin, (1) a high rising tone, which is similar to the upper rising tone (35:), and (2) a high level tone, similar to the upper even tone (55:). The results indicate that most of the cases of Bian-yin belong to the first category and that they all appear in nouns. Phonological processes are generalisations about the child's substitutions. They are natural tendencies the child uses to simplify adult target sounds. The results of this study show that there are processes that are shared by all the subjects, and others that are not. For example, all the subjects have the process of tensing vowels. This suggests that it is a common process among the Cantonese-speaking children. In addition to looking at the Cantonese produced by the subjects, this study also examines the English loanwords in Wai's speech. Her loan-words fit the rules given by H.N. Cheung (1972). For example, the English /\/ becomes a /p/ in the loan-word, and a vowel /i/ is often added to the loan-word for the English word that ends with a /s/, forming a new syllable. Some loan-words in Wai's speech are [npm1pa2] 'number', and [tsu'si4] 'juice'. Although English words and phrases occupied only a very small part of the vocabulary in Wai's speech, I look at all the English words and phrases that she used. Based on S.M. Tse (1978), the results indicate that Wai's pronunciation errors in English consonants are very similar to those made by the Cantonese adults who are learning English as a second language. The study also compares the phonological systems of English and Cantonese and argues that the phonology of Cantonese is easier to acquire than that of English. Two reasons are put forth to support the argument. They are (1) the more complex system in the English consonants and (2) the assistance of the tonal system in the acquisition of Cantonese. Moreover, I compare the results of the proportions of matches between the Cantonese subjects in this study and the English subjects based on Ingram (1981). The results show that the Cantonese subjects performed better in matching the adult models. Finally, this study moves from the analysis of the child's phonology to the consideration of the influence of the different dialects of the parents' speech on the child. This is a topic that has seldom been emphasized by other investigators. The speech of Wai's parents differs in the use of the /l/ and /n/ initial syllables. The father's dialect distinguishes both /!/ and /n/ initials, while that of the mother has all the /n/s replaced by /1/s. I examined Wai's use of [1] and [n] in initial syllables that required /l/ and /n/ in the father's dialect to see how these differences in the input language affected her production. First of all, [1] and [n] free varied throughout all sessions for both /l/ and /n/ initials. During this time, Wai went through two periods of development. The first period is characterized by a preference for nasalization, lasting from 1;7(14) to 2;0(27). The second period is characterized by the preference for lateralization, and extended approximately from age 2; 1(24) and up. These data show that Wai considered [1] and [n] to be allophones of a single phoneme. At the end of the study, however, Wai was on the verge of adopting theramother's dialect. / Arts, Faculty of / Linguistics, Department of / Graduate

Neste estudo, investigamos a hipótese de que fenômenos fonológicos que ocorrem com certos elementos morfossintáticos, como palavras compostas, nomes próprios e siglas, em português brasileiro (doravante, PB) dão evidência para a proposição do domínio prosódico denominado Grupo de Palavras Prosódicas (PWG Prosodic Word Group [Vigário, 2007, 2010]), o qual se situa na hierarquia prosódica entre a Palavra Prosódica (PWd) e o Sintagma Fonológico (PPh). Esse mesmo nível tem sido denominado grupo clítico (C) (Nespor & Vogel, 1986, 2007; Hayes, 1989) e descrito como o domínio prosódico que compreende as palavras compostas e as palavras lexicais acentuadas eventualmente acompanhadas por clíticos. Contudo, não há consenso absoluto quanto à associação de clíticos a palavras lexicais acentuadas implicar realmente a existência de um nível prosódico distinto, uma vez que esse tipo de associação pode variar: em várias línguas já foi observada a associação de clíticos a níveis superiores e inferiores ao da PWd. Cabe ainda observar que palavras compostas e palavras lexicais acentuadas acompanhadas por clíticos não são os únicos elementos morfossintáticos do PB cujo domínio prosódico em que são mapeados consiste no nível da hierarquia prosódica situado entre PWd e PPh. Para a nossa investigação, analisamos um corpus de PB no sentido de determinarmos de que forma a prosodização desses elementos em PWGs se difere ou se assemelha à prosodização de PWds e PPhs. Com esse objetivo, procuramos contrastes entre a a prosodização de elementos que, por hipótese, tenham PWG como domínio e a prosodização de elementos que tenham PPh como domínio. Ademais, procuramos, em contexto de foco contrastivo, contrastes entre o comportamento de PPhs e PWGs ramificados. Com relação à marcação de foco contrastivo em PWG, observamos que, tanto em PPhs ramificados como em palavras compostas e nomes próprios, parece não haver possibilidade de ocorrência de proeminência fonológica de foco em uma PWd diferente da PWd núcleo da focalização. Para siglas, existe essa possibilidade, ainda que o fenômeno não seja frequente. Com relação ao bloqueio de SVE pelo acento de PWG em diferentes estruturas prosódicas, não observamos comportamentos sistematicamente distintos entre PWGs e PPhs ramificados com o mesmo número de PWds em estruturas frasais análogas. Por outro lado, observamos que o acento de PWG tem influência na ocorrência de SVE mesmo quando não coincide com o acento de PPh. Portanto, de acordo com os resultados que obtivemos, o PWG se mostra relevante para a descrição prosódica do PB. / In this study, we investigate the hypothesis that phonological phenomena that occur with certain morphosyntactic elements, such as compound words, proper names and acronyms, in Brazilian Portuguese (henceforth, PB) give evidence to the proposition of the prosodic domain denominated Prosodic Word Group (PWG [Vigário, 2007, 2010]), which, within the prosodic hierarchy, is located between the Prosodic Word (PWd) and the Phonological Phrase (PPh). This same level has been called Clitic Group (C) (Nespor & Vogel, 1986, 2007; Hayes, 1989) and described as the domain which includes the compound words and the stressed lexical words eventually accompanied by clitics. However, there is not consensus as to if the association of clitics to stressed lexical words really implies the existence of a distinct prosodic level, as this kind of association can vary: in many languages the association of clitics has already been observed to levels above and below the PWd. It is also important to say that PWds and stressed lexical words accompanied by clitics are not the only morphosyntactic elements in PB whose prosodic domain in which they are mapped is the level of the prosodic hierarchy located between the PWd and the PPh. For our investigation, we analyzed a corpus of PB with the objective of determining how the prosodization of these elements in PWGs differs or resembles the prosodization of PWds and PPhs. With this objective, we have looked for contrasts between the prosodization of elements which hypothetically have the PWG as domain and the prosodization of elements which hypothetically have the PPh as domain. Moreover, we have looked for contrasts between the behavior of branching PPhs and PWGs in context of contrastive focus. As to the contrastive focus in PWG, we have observed that, both in branching PPhs, compound words and proper names, there seems to be no possibility of occurrence of phonological prominence of focus in a PWd different from the PWd that is the nucleus of the focalization. For acronyms, there is this possibility, despite the fact that this phenomenon is not frequent. As to the role of the PWG stress in blocking the phenomenon external vowel sandhi (SVE) in different prosodic structures, we have not observed systematically distinct behaviors between branching PWGs and PPhs with the same number of PWds in analogous phrasal structures. On the other hand, we have observed that the PWG stress does influence the occurrence of SVE even when it does not coincide with the PPh stress. Therefore, according to the results that we have obtained, the PWG seems relevant to the prosodic description of PB.

Neste estudo, investigamos a hipótese de que fenômenos fonológicos que ocorrem com certos elementos morfossintáticos, como palavras compostas, nomes próprios e siglas, em português brasileiro (doravante, PB) dão evidência para a proposição do domínio prosódico denominado Grupo de Palavras Prosódicas (PWG Prosodic Word Group [Vigário, 2007, 2010]), o qual se situa na hierarquia prosódica entre a Palavra Prosódica (PWd) e o Sintagma Fonológico (PPh). Esse mesmo nível tem sido denominado grupo clítico (C) (Nespor & Vogel, 1986, 2007; Hayes, 1989) e descrito como o domínio prosódico que compreende as palavras compostas e as palavras lexicais acentuadas eventualmente acompanhadas por clíticos. Contudo, não há consenso absoluto quanto à associação de clíticos a palavras lexicais acentuadas implicar realmente a existência de um nível prosódico distinto, uma vez que esse tipo de associação pode variar: em várias línguas já foi observada a associação de clíticos a níveis superiores e inferiores ao da PWd. Cabe ainda observar que palavras compostas e palavras lexicais acentuadas acompanhadas por clíticos não são os únicos elementos morfossintáticos do PB cujo domínio prosódico em que são mapeados consiste no nível da hierarquia prosódica situado entre PWd e PPh. Para a nossa investigação, analisamos um corpus de PB no sentido de determinarmos de que forma a prosodização desses elementos em PWGs se difere ou se assemelha à prosodização de PWds e PPhs. Com esse objetivo, procuramos contrastes entre a a prosodização de elementos que, por hipótese, tenham PWG como domínio e a prosodização de elementos que tenham PPh como domínio. Ademais, procuramos, em contexto de foco contrastivo, contrastes entre o comportamento de PPhs e PWGs ramificados. Com relação à marcação de foco contrastivo em PWG, observamos que, tanto em PPhs ramificados como em palavras compostas e nomes próprios, parece não haver possibilidade de ocorrência de proeminência fonológica de foco em uma PWd diferente da PWd núcleo da focalização. Para siglas, existe essa possibilidade, ainda que o fenômeno não seja frequente. Com relação ao bloqueio de SVE pelo acento de PWG em diferentes estruturas prosódicas, não observamos comportamentos sistematicamente distintos entre PWGs e PPhs ramificados com o mesmo número de PWds em estruturas frasais análogas. Por outro lado, observamos que o acento de PWG tem influência na ocorrência de SVE mesmo quando não coincide com o acento de PPh. Portanto, de acordo com os resultados que obtivemos, o PWG se mostra relevante para a descrição prosódica do PB. / In this study, we investigate the hypothesis that phonological phenomena that occur with certain morphosyntactic elements, such as compound words, proper names and acronyms, in Brazilian Portuguese (henceforth, PB) give evidence to the proposition of the prosodic domain denominated Prosodic Word Group (PWG [Vigário, 2007, 2010]), which, within the prosodic hierarchy, is located between the Prosodic Word (PWd) and the Phonological Phrase (PPh). This same level has been called Clitic Group (C) (Nespor & Vogel, 1986, 2007; Hayes, 1989) and described as the domain which includes the compound words and the stressed lexical words eventually accompanied by clitics. However, there is not consensus as to if the association of clitics to stressed lexical words really implies the existence of a distinct prosodic level, as this kind of association can vary: in many languages the association of clitics has already been observed to levels above and below the PWd. It is also important to say that PWds and stressed lexical words accompanied by clitics are not the only morphosyntactic elements in PB whose prosodic domain in which they are mapped is the level of the prosodic hierarchy located between the PWd and the PPh. For our investigation, we analyzed a corpus of PB with the objective of determining how the prosodization of these elements in PWGs differs or resembles the prosodization of PWds and PPhs. With this objective, we have looked for contrasts between the prosodization of elements which hypothetically have the PWG as domain and the prosodization of elements which hypothetically have the PPh as domain. Moreover, we have looked for contrasts between the behavior of branching PPhs and PWGs in context of contrastive focus. As to the contrastive focus in PWG, we have observed that, both in branching PPhs, compound words and proper names, there seems to be no possibility of occurrence of phonological prominence of focus in a PWd different from the PWd that is the nucleus of the focalization. For acronyms, there is this possibility, despite the fact that this phenomenon is not frequent. As to the role of the PWG stress in blocking the phenomenon external vowel sandhi (SVE) in different prosodic structures, we have not observed systematically distinct behaviors between branching PWGs and PPhs with the same number of PWds in analogous phrasal structures. On the other hand, we have observed that the PWG stress does influence the occurrence of SVE even when it does not coincide with the PPh stress. Therefore, according to the results that we have obtained, the PWG seems relevant to the prosodic description of PB.