A carregar...
Publicação
The acquisition of control in european portuguese complement clauses
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 131.34 KB | Adobe PDF | |||
| 985.67 KB | Adobe PDF | |||
| 197.54 KB | Adobe PDF | |||
| 50.48 KB | Adobe PDF |
Autores
Orientador(es)
Resumo(s)
Este estudo debruça-se sobre o desenvolvimento da gramática de controlo
em crianças monolingues falantes de Português Europeu (PE). Pretende-se avaliar a
compreensão do sujeito foneticamente nulo em completivas infinitivas por crianças
entre os 3 e os 5 anos. Assim, foi testada a sua escolha de antecedente em
completivas infinitivas, um contexto de controlo obrigatório, e em sujeitos
oracionais de infinitivo não-flexionado, um contexto de controlo não-obrigatório.
É geralmente aceite que as crianças têm conhecimento da gramática de
controlo em completivas (à exceção de completivas de promise “prometer”), embora
as crianças em estádios muito iniciais possam ter leituras de controlo nãoobrigatório
de sujeitos nulos obrigatoriamente controlados (McDaniel & Cairns,
1990b; McDaniel et al., 1991; Landau & Thornton, 2011). Como explicação,
McDaniel et al. (1991) sugerem que estas leituras se devem a uma representação
não-adulta de estruturas completivas, e que a interpretação de PRO nestes
contextos é feita através de estratégias interpretativas lineares. Em alternativa,
Sherman e Lust (1986, 1993) sugerem que o conhecimento da sintaxe de controlo é
contínuo, e que os efeitos de desenvolvimento se devem à aquisição lexical e à
integração do conhecimento lexical com o conhecimento sintático. De acordo com
esta proposta, as crianças não usam estratégias interpretativas lineares para atribuir
um antecedente a PRO. A sua interpretação deste elemento é baseada no seu
conhecimento gramatical ao longo de todo o seu percurso de desenvolvimento.
Uma tarefa de compreensão, baseada em McDaniel e Cairns (1990a, b), foi
desenvolvida para recolher juízos de referência e aplicada a 64 crianças e 20 adultos.
A tarefa reúne quatro condições, duas das quais estão divididas em duas
subcondições: (1) controlo de sujeito, com (1a) verbos transitivos (querer, conseguir) e
(1b) verbos ditransitivos (prometer), (2) controlo de objeto, com (2a) objetos diretos
(ensinar a, proibir de, pôr a) e (2b) objetos indiretos (dizer para), (3) sujeitos oracionais
(aborrecer, assustar), e (4) casos de interpretação pragmaticamente determinada (pedir
para).
Deste modo, a tarefa experimental testa contextos de controlo obrigatório
(Condições 1, 2 e 4) e um contexto de controlo não-obrigatório (Condição 3). Isto
permitir-nos-á testar se as crianças entre os 3 e os 5 anos têm interpretações de
PRO baseadas no conhecimento da gramática de controlo. Além disso, a Condição
4 irá permitir-nos avaliar qual a interpretação preferencial das crianças e dos
adultos, uma vez que ambos os potenciais antecedentes na matriz são opções
gramaticais e os contextos usados na tarefa são neutros quanto ao antecedente.
As questões de investigação que guiaram este projeto são: 1) as crianças
falantes de PE em estádios iniciais do desenvolvimento têm interpretações de PRO
baseadas no conhecimento gramatical? e 2) as crianças falantes de PE são capazes de
distinguir entre os diferentes contextos em que PRO ocorre e entre as diferentes
propriedades referenciais deste elemento nulo nesses contextos?
Foi colocada a hipótese de que se houver um estádio inicial em que PRO
tem interpretação livre, tal como foi sugerido por Hsu et al. (1989), McDaniel &
Cairns (1990b), McDaniel et al. (1991), e Eisenberg & Cairns (1994), as crianças irão
aceitar um DP sujeito, um DP objeto ou uma personagem não mencionada na frase
como o controlador de PRO, independentemente da estrutura em que este
elemento ocorra (nomeadamente em contextos de controlo obrigatório e de
controlo não-obrigatório). Se, por outro lado, a escolha de antecedente em
completivas de controlo por crianças entre os 3 e os 5 anos for restringida pelo seu
conhecimento gramatical, o seu comportamento poderá ser explicado por três
hipóteses. De acordo com a Single Argument Structure Hypothesis (SASH: Santos,
Gonçalves & Hyams 2014, in prep.), as crianças analisam os verbos de controlo
ditransitivos como tendo apenas um argumento interno proposicional, pelo que
deverão 1) ter uma preferência marcada pelo controlo de objeto em estruturas com
dois argumentos internos, 2) ter diferentes proporções de controlo de objeto com
diferentes verbos matriz ditransitivos, e 3) evidenciar a reanálise da estrutura
argumental dos verbos de controlo ditransitivos. Por outro lado, se as crianças
tiverem as mesmas proporções de controlo de objeto com todos os verbos
ditransitivos, o seu comportamento poderá ser explicado por efeitos de intervenção
(Argument Intervention Hypothesis, AIH: Orfitelli 2012a, b). Estendendo esta hipótese
a contextos de controlo, e assumindo uma análise de controlo como movimento
(Hornstein 1999), pode considerar-se que o DP objeto irá intervir no controlo de
sujeito com verbos como prometer. Por último, e dado que o PE é uma língua prodrop,
a saliência do sujeito mais alto como um potencial antecedente para sujeitos
nulos encaixados poderá afetar a escolha de antecedente em contextos de controlo
(Montalbetti 1984), pelo que as crianças poderão dar respostas de controlo de
sujeito em contextos de controlo de objeto. Relativamente à segunda questão de
investigação, se as crianças forem de facto capazes de distinguir entre os diferentes
contextos em que PRO ocorre e entre as diferentes propriedades referenciais que
este elemento nulo tem nesses contextos, as respostas de controlo por um
antecedente não mencionado na frase ficarão restringidas a contextos de controlo
não-obrigatório (nomeadamente a sujeitos oracionais).
Os dados experimentais indicam que as crianças conseguem identificar
contextos de controlo obrigatório e de controlo não-obrigatório, uma vez que
escolhem um antecedente não mencionado na frase apenas nos sujeitos oracionais.
Na condição de controlo de sujeito, as crianças responderam de acordo com a
gramática adulta apenas com os verbos transitivos (Condição 1a), isto é, na ausência
de um DP objeto na matriz. Com o verbo prometer, as respostas de controlo de
sujeito das crianças ficaram muito abaixo dos níveis adultos, o que pode ser
resultado de efeitos de intervenção (Friedmann, Belletti & Rizzi 2009; Orfitelli
2012a, b para efeitos semelhantes com movimento-A), se o controlo for analisado
como movimento (Hornstein, 1999). No entanto, estes efeitos também podem ser
resultado de uma análise não-adulta da estrutura argumental dos verbos com dois
argumentos internos (um DP objeto e um argumento oracional), isto é, estes efeitos
podem dever-se à Single Argument Structure Hypothesis (SASH: Santos, Gonçalves &
Hyams 2014, in prep.). Este estudo demonstrou que as crianças falantes de PE entre
os 3 e os 5 anos de idade analisam os verbos de controlo de objeto como tendo
apenas um argumento interno, em que o DP objeto da gramática adulta constitui
um sujeito da infinitiva. Isto dá origem a produções agramaticais com preposições
mal colocadas e infinitivos flexionados em contextos de infinitivo não-flexionado.
Se o DP objeto selecionado por um verbo de controlo ditransitivo for tomado como
o sujeito de uma oração não-finita, a criança terá leituras de controlo de objeto com
verbos de controlo de objeto e com verbos como prometer. Isto é, em estruturas de
controlo de objeto a criança terá leituras aparentemente adultas com uma análise
não-adulta da estrutura completiva. Esta hipótese também prediz melhores
resultados na compreensão de controlo de objeto com verbos que apresentam mais
casos deste tipo de análise não-adulta nos dados de produção provocada de Santos,
Gonçalves & Hyams (2014, in prep.), nomeadamente com proibir de. Os dados
experimentais demonstram que esta predição é confirmada: as crianças têm
melhores resultados com proibir de do que com ensinar a. Além disso, as crianças (à
semelhança dos adultos) têm uma forte preferência por controlo de objeto com
pedir para, o que poderá igualmente dever-se à SASH, em estádios iniciais.
Abstract: This study addresses the development of Control in monolingual children acquiring European Portuguese (EP). It aims at assessing children’s comprehension of the phonetically null infinitival subject in Obligatory Control (OC) contexts, as well as in sentential subjects, an NOC context. Hence, children’s choice of antecedent in infinitival complement clauses and sentential subjects was tested using a comprehension task based on McDaniel and Cairns (1990a, b). It is generally assumed that children have early knowledge of control in complement clauses (with the exception of complements of promise), although very young children may have non-obligatory control readings of obligatorily controlled null embedded subjects (McDaniel & Cairns, 1990b; McDaniel et al., 1991; Landau & Thornton, 2011). As an explanation, it has been suggested that the syntactic representation of complement clauses is non target-like at early stages of development (McDaniel et al., 1991), and that non-adult answers may follow from linear interpretative strategies (Hsu & Cairns, 1990 apud McDaniel & Cairns 1990a). Alternatively it has been suggested that knowledge of the syntax of control is continuous, and developmental effects follow from lexical acquisition and from the integration of lexical knowledge with syntactic knowledge (Sherman & Lust 1986, 1993). According to this proposal, children show some knowledge of the grammar of control, as they are able to identify control contexts, and they do not use linear interpretative strategies to establish PRO’s reference. Their interpretation of PRO is grammatically based at all times throughout their development. A comprehension task, based on McDaniel and Cairns (1990a, b), was developed to elicit referential judgments and applied to 64 children aged 3 to 5 and 20 adults. It comprises four test conditions, two of which are divided into two subconditions: (1) subject control, with (1a) transitive verbs (querer “want”, conseguir “manage to”) and (1b) ditransitive verbs (prometer “promise”), (2) object control, with (2a) direct objects (ensinar a “teach”, proibir de “forbid”, pôr a “put to”) and (2b) indirect objects (dizer para “tell”), (3) sentential subjects (aborrecer “bother”, assustar “scare”), and (4) cases of pragmatically determined interpretation (pedir para ‘ask’). Hence, this task tests both obligatory control (OC) contexts (Conditions 1, 2 and 4) and a non-obligatory control (NOC) context (Condition 3). This will allow us to assess whether children have grammatically based interpretations of PRO, and whether children have a stage of free interpretation of PRO. In addition, Condition 4 will allow us to assess children’s and adults’ preferred interpretation, given that both sentence-internal potential antecedents are grammatical options and the contexts used in the task were designed to be neutral. Our research questions are: 1) do Portuguese-speaking children show evidence of grammatically based interpretations of PRO at early stages? and 2) are Portuguese-speaking children able to distinguish between different PRO-contexts and the subsequent varying referential properties of this null element? It was hypothesized that if there is an early stage of free interpretation of PRO, as suggested by Hsu et al. (1989), McDaniel & Cairns (1990b), McDaniel et al. (1991), and Eisenberg & Cairns (1994), children will accept a subject DP, an object DP or a third-character as the controller of PRO, regardless of the structure it occurs in (namely in both obligatory control and non-obligatory control configurations). If, on the other hand, children’s choice of antecedent in control complements is grammatically constrained, children’s behavior may be explained by three hypotheses. According to the Single Argument Structure Hypothesis (SASH: Santos, Gonçalves & Hyams 2014, in prep.), the acquisition of control is contingent on the acquisition of control verbs’ argument structure: children initially misanalyze ditransitive control verbs as taking a single propositional complement, and they must subsequently reanalyze the argument structure of these predicates. Consequently, young children should 1) show a strong preference for object control in structures with two internal arguments, 2) evidence different rates of object control responses with different matrix verbs, and 3) show evidence of reanalysis of ditransitive control verbs’ argument structure. Conversely, if children prefer object control with all ditransitive control verbs equally, their performance may be explained by Orfitelli’s (2012a, b) Argument Intervention Hypothesis (AIH), an explanatory proposal that was made for the delayed acquisition of raising-to-subject with experiencer arguments in English, with verbs such as seem and appear. Extending this hypothesis to control structures with two internal arguments, and assuming a movement account of control (Hornstein 1999), an object DP would act as an intervener for subject control with promise-type verbs. Finally, given that EP is a pro-drop language, the saliency of the higher DP subject as an antecedent for embedded null subjects may affect the choice of antecedent in control contexts (Montalbetti 1984), and children may show subject control interpretations in object control contexts. Concerning the second research question, if children are in fact able to distinguish between the different contexts in which PRO occurs and the subsequent varying referential properties of this null element, third character responses will be restricted to non-obligatory control contexts (in particular to sentential subjects). The experimental results indicate that children are able to distinguish between obligatory control and non-obligatory control contexts, given that they choose an unmentioned character as the antecedent only in sentential subjects. Children performed at adult level with subject control only with transitive verbs, that is, in the absence of a DP object in the matrix clause. In the case of prometer ‘promise’ children performed far below adult level, a fact that could result from intervention effects (Friedmann, Belletti & Rizzi 2009; Orfitelli 2012a, b for similar effects on A-movement) if control is analyzed as movement (Hornstein, 1999). However, these effects may also result from a misanalysis of the argument structure of verbs with two internal arguments (a DP object and a clausal argument), that is, these effects may be due to the Single Argument Structure Hypothesis (SASH: Santos, Gonçalves & Hyams 2014, in prep.). This study showed that Portuguesespeaking children between 3 and 5 years of age misanalyse object control verbs as verbs taking a single internal (propositional) argument: they take the DP object to be the subject of the infinitive, which results in non-target like utterances with misplaced prepositions (the preposition is placed to the right of the verb, instead of being placed to the right of the target grammar DP object) and inflected infinitives in uninflected infinite contexts. The inflected infinitive, then, is able to license the target grammar DP object as an infinitival subject. If the DP object selected by either an object control verb or a promise-type verb is taken to be the subject of the non-finite complement clause, what appears to be an object control reading can come for free, resulting in seemingly target comprehension of object control and non-target comprehension of subject control with promise. This hypothesis also predicts better comprehension results with the object control verbs that presented more cases of the type of misanalysis identified by Santos, Gonçalves & Hyams (2014, in prep.), in their elicited production task, namely proibir de ‘forbid’. Our experimental data shows that this prediction is confirmed: children have better comprehension results with proibir de “forbid” than with ensinar a “teach”. Furthermore, both children and adults show a strong preference for object control with pedir para “ask”, which may also be due to the Single Argument Structure Hypothesis, at the initial stages of development.
Abstract: This study addresses the development of Control in monolingual children acquiring European Portuguese (EP). It aims at assessing children’s comprehension of the phonetically null infinitival subject in Obligatory Control (OC) contexts, as well as in sentential subjects, an NOC context. Hence, children’s choice of antecedent in infinitival complement clauses and sentential subjects was tested using a comprehension task based on McDaniel and Cairns (1990a, b). It is generally assumed that children have early knowledge of control in complement clauses (with the exception of complements of promise), although very young children may have non-obligatory control readings of obligatorily controlled null embedded subjects (McDaniel & Cairns, 1990b; McDaniel et al., 1991; Landau & Thornton, 2011). As an explanation, it has been suggested that the syntactic representation of complement clauses is non target-like at early stages of development (McDaniel et al., 1991), and that non-adult answers may follow from linear interpretative strategies (Hsu & Cairns, 1990 apud McDaniel & Cairns 1990a). Alternatively it has been suggested that knowledge of the syntax of control is continuous, and developmental effects follow from lexical acquisition and from the integration of lexical knowledge with syntactic knowledge (Sherman & Lust 1986, 1993). According to this proposal, children show some knowledge of the grammar of control, as they are able to identify control contexts, and they do not use linear interpretative strategies to establish PRO’s reference. Their interpretation of PRO is grammatically based at all times throughout their development. A comprehension task, based on McDaniel and Cairns (1990a, b), was developed to elicit referential judgments and applied to 64 children aged 3 to 5 and 20 adults. It comprises four test conditions, two of which are divided into two subconditions: (1) subject control, with (1a) transitive verbs (querer “want”, conseguir “manage to”) and (1b) ditransitive verbs (prometer “promise”), (2) object control, with (2a) direct objects (ensinar a “teach”, proibir de “forbid”, pôr a “put to”) and (2b) indirect objects (dizer para “tell”), (3) sentential subjects (aborrecer “bother”, assustar “scare”), and (4) cases of pragmatically determined interpretation (pedir para ‘ask’). Hence, this task tests both obligatory control (OC) contexts (Conditions 1, 2 and 4) and a non-obligatory control (NOC) context (Condition 3). This will allow us to assess whether children have grammatically based interpretations of PRO, and whether children have a stage of free interpretation of PRO. In addition, Condition 4 will allow us to assess children’s and adults’ preferred interpretation, given that both sentence-internal potential antecedents are grammatical options and the contexts used in the task were designed to be neutral. Our research questions are: 1) do Portuguese-speaking children show evidence of grammatically based interpretations of PRO at early stages? and 2) are Portuguese-speaking children able to distinguish between different PRO-contexts and the subsequent varying referential properties of this null element? It was hypothesized that if there is an early stage of free interpretation of PRO, as suggested by Hsu et al. (1989), McDaniel & Cairns (1990b), McDaniel et al. (1991), and Eisenberg & Cairns (1994), children will accept a subject DP, an object DP or a third-character as the controller of PRO, regardless of the structure it occurs in (namely in both obligatory control and non-obligatory control configurations). If, on the other hand, children’s choice of antecedent in control complements is grammatically constrained, children’s behavior may be explained by three hypotheses. According to the Single Argument Structure Hypothesis (SASH: Santos, Gonçalves & Hyams 2014, in prep.), the acquisition of control is contingent on the acquisition of control verbs’ argument structure: children initially misanalyze ditransitive control verbs as taking a single propositional complement, and they must subsequently reanalyze the argument structure of these predicates. Consequently, young children should 1) show a strong preference for object control in structures with two internal arguments, 2) evidence different rates of object control responses with different matrix verbs, and 3) show evidence of reanalysis of ditransitive control verbs’ argument structure. Conversely, if children prefer object control with all ditransitive control verbs equally, their performance may be explained by Orfitelli’s (2012a, b) Argument Intervention Hypothesis (AIH), an explanatory proposal that was made for the delayed acquisition of raising-to-subject with experiencer arguments in English, with verbs such as seem and appear. Extending this hypothesis to control structures with two internal arguments, and assuming a movement account of control (Hornstein 1999), an object DP would act as an intervener for subject control with promise-type verbs. Finally, given that EP is a pro-drop language, the saliency of the higher DP subject as an antecedent for embedded null subjects may affect the choice of antecedent in control contexts (Montalbetti 1984), and children may show subject control interpretations in object control contexts. Concerning the second research question, if children are in fact able to distinguish between the different contexts in which PRO occurs and the subsequent varying referential properties of this null element, third character responses will be restricted to non-obligatory control contexts (in particular to sentential subjects). The experimental results indicate that children are able to distinguish between obligatory control and non-obligatory control contexts, given that they choose an unmentioned character as the antecedent only in sentential subjects. Children performed at adult level with subject control only with transitive verbs, that is, in the absence of a DP object in the matrix clause. In the case of prometer ‘promise’ children performed far below adult level, a fact that could result from intervention effects (Friedmann, Belletti & Rizzi 2009; Orfitelli 2012a, b for similar effects on A-movement) if control is analyzed as movement (Hornstein, 1999). However, these effects may also result from a misanalysis of the argument structure of verbs with two internal arguments (a DP object and a clausal argument), that is, these effects may be due to the Single Argument Structure Hypothesis (SASH: Santos, Gonçalves & Hyams 2014, in prep.). This study showed that Portuguesespeaking children between 3 and 5 years of age misanalyse object control verbs as verbs taking a single internal (propositional) argument: they take the DP object to be the subject of the infinitive, which results in non-target like utterances with misplaced prepositions (the preposition is placed to the right of the verb, instead of being placed to the right of the target grammar DP object) and inflected infinitives in uninflected infinite contexts. The inflected infinitive, then, is able to license the target grammar DP object as an infinitival subject. If the DP object selected by either an object control verb or a promise-type verb is taken to be the subject of the non-finite complement clause, what appears to be an object control reading can come for free, resulting in seemingly target comprehension of object control and non-target comprehension of subject control with promise. This hypothesis also predicts better comprehension results with the object control verbs that presented more cases of the type of misanalysis identified by Santos, Gonçalves & Hyams (2014, in prep.), in their elicited production task, namely proibir de ‘forbid’. Our experimental data shows that this prediction is confirmed: children have better comprehension results with proibir de “forbid” than with ensinar a “teach”. Furthermore, both children and adults show a strong preference for object control with pedir para “ask”, which may also be due to the Single Argument Structure Hypothesis, at the initial stages of development.
Descrição
Palavras-chave
Língua portuguesa - Complementos (Linguística) Língua portuguesa - Sintaxe Controlo (Linguística) Teses de mestrado - 2014