| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 674.59 KB | Adobe PDF |
Autores
Orientador(es)
Resumo(s)
A presente dissertação tem como objetivo investigar a aquisição, por
falantes monolingues de Português Europeu, da compreensão de sujeitos
nulos em orações completivas selecionadas pelos verbos querer e dizer. Deste
modo, pretende-se aceder à interpretação de um sujeito nulo em orações
completivas de conjuntivo por crianças em idade pré-escolar (3 – 5 anos) e
escolar (6 – 8 anos), verificando se as crianças respeitam a obviação
referencial obrigatória no caso de complementos de conjuntivo e ainda se
interpretam os contextos de conjuntivo selecionados pelo verbo dizer como
contextos de controlo obrigatório.
Observou-se que os verbos querer e dizer poderão desencadear
contextos de obviação diferentes. O verbo querer representa um caso de
obviação referencial canónica, pois nestes contextos, o sujeito nulo encaixado
é obrigatoriamente disjunto do sujeito da oração matriz. O verbo dizer parece
desencadear um tipo de obviação diferente, pois ao haver a obrigatoriedade
de correferência entre o sujeito nulo da oração complemento e o objeto
indireto da oração matriz, verifica-se uma semelhança de comportamento
com os contextos de controlo obrigatório em orações não finitas. Deste modo,
este tipo de estruturas com o verbo dizer poderão aproximar-se de controlo
finito, do tipo dos já descritos por Landau (2004).
A obviação referencial é um fenómeno para o qual se tem procurado
encontrar uma explicação e, por isso, várias hipóteses têm surgido ao longo
do tempo. Por exemplo, para Picallo (1985), o domínio de ligação do
pronome encaixado, no caso de um complemento de conjuntivo, é toda a
frase e, por essa razão, há obrigatoriedade de referência disjunta nestes
contextos, caso contrário, verificar-se-ia uma violação do Princípio B. Já
Farkas (1992) e Horstein & San Martin (2000) afirmam que a obviação
referencial surge de uma competição entre os modos infinitivo e conjuntivo.
Finalmente, Kempchinksy (2009) sugere que complementos no modo
conjuntivo lexicalmente selecionado comportam um operador quasiimperative
e o papel que o operador desempenha irá culminar na obviação
referencial.
Devido à complexidade do tema em questão, a sua aquisição é
caracterizada na literatura como um fenómeno de aquisição tardia. Esta
hipótese é assumida por Padilla (1990) e Goodluck et al. (2001) para o
espanhol, por exemplo, e, até certo ponto, por Avrutin & Wexler (2000) para
o russo. No caso do Português Europeu não há muitos estudos que
investiguem a aquisição da obviação referencial como tema central, no entanto, Ambulate (2008) e Silva (2015) mostram que as crianças portuguesas
entre os 3 e os 6 anos exibem bastantes dificuldades na interpretação de um
sujeito nulo de completivas no modo conjuntivo. Com efeito, as crianças
testadas pelas autoras preferem uma leitura de correferência entre o sujeito
da oração matriz e o sujeito da oração complemento.
Assim, foi desenhada uma tarefa de compreensão baseada em
McDaniel & Cairns (1990), uma tarefa de juízo de referência que foi aplicada
a 90 crianças entre os 3 e os 8 anos e também a 30 adultos. A tarefa
experimental é constituída por 8 condições de teste. Testa-se a interpretação
de sujeitos nulos encaixados em complementos selecionados pelo verbo dizer,
sendo que se consideram casos de sujeitos nulos encaixados com dizer –
declarativo (modo indicativo) e dizer - declarativo de ordem (modo conjuntivo e
infinitivo). Testa-se ainda a interpretação de sujeitos nulos encaixados em
complementos no modo infinitivo e no modo conjuntivo selecionados pelo
verbo querer. As condições com dizer – declarativo testam a possibilidade de
leituras de correferência preferencial entre sujeito matriz e sujeito encaixado
e as condições com dizer - declarativo de ordem testam contextos de controlo
obrigatório e de obviação não canónica. Em ambos os casos são testados
contextos com objeto indireto explícito e implícito. Finalmente, com o verbo
querer são testados contextos de controlo de sujeito (complementos
infinitivos) e contextos de obviação referencial canónica (complementos
finitos, de conjuntivo).
Os resultados obtidos permitem concluir que as crianças, em
estruturas de controlo no modo infinitivo, apresentam melhores resultados
em estruturas de controlo de sujeito (querer), nas quais igualam os adultos
desde os estádios iniciais, do que em estruturas de controlo de objeto indireto
(dizer – declarativo de ordem).
Verificou-se ainda que, em estruturas com objeto indireto explícito,
quer no modo infinitivo, quer no modo conjuntivo, os resultados obtidos
pelas crianças aproximam-se dos resultados obtidos pelos adultos. Foi
notória a influência da introdução de um antecedente lexicalmente realizado,
que facilita a leitura que as crianças fazem de um sujeito nulo encaixado. Pelo
contrário, a omissão do argumento interno da oração matriz levanta
bastantes dificuldades às crianças no caso de complementos em ambos os
modos, levando-as a um desempenho muito diferente do desempenho dos
adultos.
Em complementos no modo indicativo (dizer – declarativo), em
contextos com objeto indireto explícito, a introdução de um antecedente
linguístico lexicalmente realizado parece confundir as crianças, levando-as a
alguma indecisão, mais notória no grupo de crianças mais novas, na escolha
do potencial antecedente do sujeito nulo encaixado. Nestes casos, verifica-se
que algumas crianças optam pela correferência do sujeito nulo com o objeto
indireto. Em contextos de objeto indireto implícito, as crianças igualam o
desempenho dos adultos, preferindo a correferência do sujeito encaixado
com o sujeito matriz. Estes dados corroboram os dados já obtidos por alguns
autores, por exemplo por Silva (2015).
Finalmente, os resultados revelaram que a obviação referencial não se
encontra adquirida até aos 8 anos, pelo que se revela um fenómeno de
aquisição bastante tardia. Verificou-se que em complementos de conjuntivo,
tanto em contextos com querer, como em contextos com dizer – declarativo de
ordem com objeto indireto implícito, as crianças tendem a escolher o sujeito
matriz como sendo correferente com o sujeito da oração complemento, pelo
que as crianças não distinguem os dois tipos de obviação propostos nesta
dissertação.
This study intends to investigate the acquisition of null subjects by monolingual European Portuguese children. The main goal is to assess the interpretation of a null subject in subjunctive complements selected by the verbs querer (want) and dizer (tell) by children between 3 and 8 years. Thus, this investigation will show if children respect obviation and if they interpret subjunctive contexts selected by dizer (tell) as obligatory control contexts. It is possible that the verbs querer (want) and dizer (tell) trigger different contexts of obviation. Querer (want) represents a case of canonic obviation since in these contexts, a disjoint reference between the null embedded subject and the matrix subject is mandatory, but no further restrictions exist on the interpretation of the embedded subject. Dizer (tell), on the other hand, seems to trigger a different type of obviation. The disjoint reference between subjects is also true in these contexts but it is only true because there is a mandatory coreference between the null embedded subject and the indirect object in the matrix clause. This means that the relation between the embedded subject and its potential antecedent is similar to that in obligatory control contexts in non-finite complements. Consequently, obviation contexts with dizer (tell) might be to a certain extent similar to contexts of finite control as observed by Landau (2004) for Hebrew and the Balkan languages. Through the years many hypothesis have been posed in an attempt to explain obviation effects. Picallo (1985), for example, claims that in subjunctive contexts the binding domain of an embedded pronoun is the whole sentence; consequently obviation occurs, otherwise there would be a violation of Principle B. Farkas (1992) and Horstein & San Martin (2000) assume that obviation is a consequence of mood competition (infinitive vs. subjunctive). Finally, Kempchinksy (2009) argues that lexically selected subjunctive complements carry a quasi-imperative operator that will culminate in obviation. In acquisition, obviation poses children some problems and is known as being a late acquisition. This is shown by Padilla (1990) and Goodluck et al. (2001) for Spanish and, to a certain extent, by Avrutin & Wexler (2000) for Russian. Regarding European Portuguese, not many studies concerning obviation have been made, however Ambulate (2008) and Silva (2015) show that children between 3 and 6 years old have great difficulties interpreting a null embedded subject in a subjunctive complement. The children tested by both authors preferred the coreference between the null embedded subject and the matrix subject. Thus, a comprehension task (Reference Judgment Task) based on McDaniel & Cairns (1990) was created and applied to 90 children aged between 3 and 8 years old and also 30 adults. The experimental task comprises 8 test conditions. We assess the interpretation of a null embedded subject in complements selected by the verb dizer (tell), including dizer – declarative and dizer - directive. In addition, we test the interpretation of a null embedded subject in subjunctive and infinitive complements selected by the verb querer (want). The conditions with dizer – declarative test the preferential coreference between the matrix and the embedded subject and conditions with dizer - directive test non-finite obligatory control and non canonic obviation. In both cases, contexts with an explicit and an implicit indirect object are tested. Finally, with the verb querer (want) contexts of subject control (non-finite complements) and canonic obviation (finite subjunctive complements) are tested. The results show that children perform differently in subject control contexts, where they equal the results of adults, and in indirect object control (directive dizer). It was clear that in explicit indirect object control structures, in infinitive and subjunctive mood, children’s results improved considerably, getting close to the adults´ results. The introduction of a lexical antecedent improves children’s interpretation of the embedded null subject. On the other hand, an implicit object raises difficulties in both moods, making children decrease their performance and showing non-target like results. In indicative complements, in explicit indirect object contexts, the introduction of a lexical antecedent also raises difficulties to children. The results show that some of the younger children are influenced by the presence of the indirect object, choosing it as the antecedent of the null embedded subject. In implicit indirect object contexts, children equal the adults’ results, showing preference for the coreference between the matrix and the null embedded subject. Finally, the results reveal that obviation is not acquired at 8 years old, confirming previous results suggesting that this is a phenomenon of late acquisition. Moreover, in subjunctive complements, contexts with the verb querer (want) and dizer – declarativo de ordem (directive dizer) are treated equally, i.e. children tend to prefer the coreference between the matrix and the null embedded subject, which means that children do not distinguish canonic and non canonic obviation.
This study intends to investigate the acquisition of null subjects by monolingual European Portuguese children. The main goal is to assess the interpretation of a null subject in subjunctive complements selected by the verbs querer (want) and dizer (tell) by children between 3 and 8 years. Thus, this investigation will show if children respect obviation and if they interpret subjunctive contexts selected by dizer (tell) as obligatory control contexts. It is possible that the verbs querer (want) and dizer (tell) trigger different contexts of obviation. Querer (want) represents a case of canonic obviation since in these contexts, a disjoint reference between the null embedded subject and the matrix subject is mandatory, but no further restrictions exist on the interpretation of the embedded subject. Dizer (tell), on the other hand, seems to trigger a different type of obviation. The disjoint reference between subjects is also true in these contexts but it is only true because there is a mandatory coreference between the null embedded subject and the indirect object in the matrix clause. This means that the relation between the embedded subject and its potential antecedent is similar to that in obligatory control contexts in non-finite complements. Consequently, obviation contexts with dizer (tell) might be to a certain extent similar to contexts of finite control as observed by Landau (2004) for Hebrew and the Balkan languages. Through the years many hypothesis have been posed in an attempt to explain obviation effects. Picallo (1985), for example, claims that in subjunctive contexts the binding domain of an embedded pronoun is the whole sentence; consequently obviation occurs, otherwise there would be a violation of Principle B. Farkas (1992) and Horstein & San Martin (2000) assume that obviation is a consequence of mood competition (infinitive vs. subjunctive). Finally, Kempchinksy (2009) argues that lexically selected subjunctive complements carry a quasi-imperative operator that will culminate in obviation. In acquisition, obviation poses children some problems and is known as being a late acquisition. This is shown by Padilla (1990) and Goodluck et al. (2001) for Spanish and, to a certain extent, by Avrutin & Wexler (2000) for Russian. Regarding European Portuguese, not many studies concerning obviation have been made, however Ambulate (2008) and Silva (2015) show that children between 3 and 6 years old have great difficulties interpreting a null embedded subject in a subjunctive complement. The children tested by both authors preferred the coreference between the null embedded subject and the matrix subject. Thus, a comprehension task (Reference Judgment Task) based on McDaniel & Cairns (1990) was created and applied to 90 children aged between 3 and 8 years old and also 30 adults. The experimental task comprises 8 test conditions. We assess the interpretation of a null embedded subject in complements selected by the verb dizer (tell), including dizer – declarative and dizer - directive. In addition, we test the interpretation of a null embedded subject in subjunctive and infinitive complements selected by the verb querer (want). The conditions with dizer – declarative test the preferential coreference between the matrix and the embedded subject and conditions with dizer - directive test non-finite obligatory control and non canonic obviation. In both cases, contexts with an explicit and an implicit indirect object are tested. Finally, with the verb querer (want) contexts of subject control (non-finite complements) and canonic obviation (finite subjunctive complements) are tested. The results show that children perform differently in subject control contexts, where they equal the results of adults, and in indirect object control (directive dizer). It was clear that in explicit indirect object control structures, in infinitive and subjunctive mood, children’s results improved considerably, getting close to the adults´ results. The introduction of a lexical antecedent improves children’s interpretation of the embedded null subject. On the other hand, an implicit object raises difficulties in both moods, making children decrease their performance and showing non-target like results. In indicative complements, in explicit indirect object contexts, the introduction of a lexical antecedent also raises difficulties to children. The results show that some of the younger children are influenced by the presence of the indirect object, choosing it as the antecedent of the null embedded subject. In implicit indirect object contexts, children equal the adults’ results, showing preference for the coreference between the matrix and the null embedded subject. Finally, the results reveal that obviation is not acquired at 8 years old, confirming previous results suggesting that this is a phenomenon of late acquisition. Moreover, in subjunctive complements, contexts with the verb querer (want) and dizer – declarativo de ordem (directive dizer) are treated equally, i.e. children tend to prefer the coreference between the matrix and the null embedded subject, which means that children do not distinguish canonic and non canonic obviation.
Descrição
Palavras-chave
Língua portuguesa - Aquisição Língua portuguesa - Sujeitos nulos Sintaxe Teoria da regência e da ligação Teses de mestrado - 2017
