Abstract
Anyone who knows a language can tell whether an answer is compatible (C) or incompatible (I) with a question. Subjects were presented with a variety of C and I question and answer pairs and allowed as much time as they required to commit each to memory. It was found that subjects took longer for the initial processing of I items than of C items, with performance on the fastest I sub-class worse than performance on the slowest C sub-class. Nevertheless, at recall subjects did worse on I items than C items, with performance on the best I sub-class worse than that on the poorest C sub-class. The most popular error on I items was such as to change these to C items. Subjects' knowledge of the linguistic constraints obtaining across a sentence boundary systematically affected performance. The task ahead is to specify how this may happen.
Get full access to this article
View all access options for this article.