• No se han encontrado resultados

La producción de datos concretos y los censos de expatriados

In sum, there is compelling evidence to support both approaches to syntax acquisition but, currently, neither approach can fully explain the pattern of children’s early syntactic knowledge. Early abstraction accounts like the theory of universal grammar, have problems explaining why children’s early speech is not adult-like; if children have innate knowledge of linguistic parameters and principles that help them to constrain the possible

hypotheses about the language that they are hearing, why do they make overgeneralisation errors in their early speech? In addition, if children have such powerful innate linguistic knowledge that affords them the potential to produce an infinite number of novel utterances, why is it that a large

proportion of their early multi-word utterances are explained by a relatively small number of lexically-based schemas (Lieven, Pine, & Baldwin, 1997)? In other words, unlike lexical constructivist accounts, which predict a close relationship between the lexicon and syntactic structure, early abstraction accounts have problems explaining the lexical specificity of children’s early speech. That is not to say that lexical constructivist accounts are without their issues. Although accounts like the usage-based theory can explain why children’s early speech tends to be built around highly frequent item-based fragments, they, like nativist accounts, struggle to explain how children are able to constrain their speech so that do not end up with an over-general

26

Chapter 1 grammar. These accounts also have difficulty explaining why children show

evidence of abstract knowledge in their language comprehension (i.e., why they are able to understand sentences with novel verbs from early in the acquisition process) (e.g., Naigles, 1990; Gertner, Fisher, & Eisengart, 2006).

To add to this, both elicited-production and preferential-looking tasks - the traditionally-preferred paradigms used to test the abstractness of

children’s syntactic representations, may contain methodological flaws that make it difficult to draw firm conclusions about the way in which children represent syntax. First, both types of paradigm use novel verbs. This is because the use of novel verbs allows a high degree of control over the language to which the children are exposed, ensuring that the speech produced or comprehended is creative and does not occur as a result of previous experience. The problem with this is that, as argued by Fisher (2002), using novel verbs in production tasks also introduces the possibility that young children will have difficulty producing these new verbs after only a few experimental sessions; the increased cognitive load (as a result of the additional memory demands when learning new verbs) during elicited-

production tasks might make it difficult for children to access the appropriate syntactic representations (Valian, Solt, & Stewart, 2009). Another

disadvantage of using novel verbs is that it is not always clear that children will have learned the verb’s meaning. For example, while the use of

preferential-looking tasks has many advantages (e.g., enabling researchers to test children at a much earlier stage of the acquisition process than elicited production tasks will allow), these tasks may not directly test whether

27

Chapter 1 children understand the meaning of verbs that they have never encountered

before. They may instead test whether or not children can tell the difference between a pair of sentences – which, arguably, might not rely on abstract syntactic knowledge. For example, in Naigles’ (1990) task, it may simply be

that for intransitive sentences, children associated the word ‘and’ with the

absence of causation, in which case, knowledge of abstract syntax might not

be necessary in determining that “The bunny and the duck are gorping” is a

non-causal event. Another problem with these types of tasks is that the time spent looking at the screens is not a fail-safe measure of whether or not children understand the sentences that they are hearing. For instance, in Gertner et al.’s (2006) study, children may have just looked longer at the correct screen because they preferred to look at the first-mentioned

character who, as the agent, is always the protagonist. Alternatively, children may look longer at the incorrect screen because they find the discrepancy between the video and the sentence more interesting (Ambridge & Rowland, 2013). Thus, while these studies might claim that young children are able to use syntactic knowledge to understand novel sentences, they do not provide conclusive evidence that this knowledge is sophisticated enough to enable them to generalise across a range of sentence types. Further still, in

production tasks, children are exposed to each novel verb many times during training sessions. Effectively, the novel verbs are primed in one structure before the child is encouraged to use it in another, making it potentially more difficult for the child to use the novel verbs in a different way.

An alternative way of adjudicating between lexical constructivist and early abstraction accounts is to use structural priming since this paradigm

28

Chapter 1 uses familiar verbs and tests for the effects of priming directly. Not only does

the structural priming paradigm obviate the need for novel stimuli by using familiar verbs, but the degree of overlap in lexical content between prime and target sentences can be controlled, allowing for the investigation of different levels of abstractness in children’s early syntactic representations. The next section reviews the developmental priming literature. First, we explain how the structural priming paradigm works. Then, we describe the effects that have been found, and discuss what these effects can tell us about children’s early syntactic representations.