Acquisition of syntax in a miniature artificial language : effects of input and instruction
The goal of the research was to discover which type of input and instruction best facilitates the acquisition of syntax in adult learners. An artificial miniature language was used to model real second language to control precisely the type of input, conditions of exposure and instruction accompanying that input. Performance of learners under four input conditions was compared and analogies were drawn between these conditions and those experienced by adult second language learners (L2 learners). 'Instructed' learners Z): like formally instructed L2 learners were systematically taught the rules of the language. 'Exposure' learners saw example sentences and were asked to search for rules, the conditions of their input analogous to that of 'naturalistic' L2 learners who receive no formal instruction but who make conscious efforts to search for rules. 'Memorisation' learners received the same input as that presented to the exposure learners but were asked to memorise the sentencesT. hey were seena s analogoust o naturalistic L2 learners who do not search for the rules and the conditions of input were modelled on those claimed to induce implicit learning. 'Cued' learners received input which contained cross-sentential cues to underlying phrasal structure. They were modelled on naturalistic learners whose input contains such cues and who make efforts to search for rules. Performance was compared on both grammaticality judgement and free production tasks. No overall superiority in performance was observed for any of the input conditions. An interaction between input type and rule complexity was evident in which the amount of information received regarding the rules related positively to performance on. the less salient, more complex rules. It was proposed that the findings could be explained in terms of a 'noticing' hypothesis, in which noticing of features is considered a pre-requisite for acquisition. Theories of second and artificial language learning which have stipulated that complex rules can only be learned implicitly were not supported.