Cultural Differences in Language Processing


University of Glasgow Department of Psychology	Johannes Gutenberg - Universität, Mainz Psychologisches Institut

About the Constituent Comparison Model

For this investigation a serial model was used whereby information from two sources (written & pictorial) is to compare and to decide whether it is true with respect to one another. Basically, a sentence and a picture is presented to subjects on a computer screen and subjects are to judge whether the sentence is true for that particular picture. Subjects have to press one of two buttons, indicating their answer (true or false), whereby their response latency is measured. Sentences used for this investigation are, for instance The squares are not red. The picture was an array of squares, all in one colour, either the mentioned one or a different colour. The propositional format for the example sentence is (NEG [RED; SQUARES]), denoting the argument SQUARES, the predicate RED, and the polarity marker NEG. In affirmative sentences the marker is AFF. Pictures are always represented affirmatively. A truth index that can have two values records the response value and is set initially to true. The comparison process starts at the inner most constituent and proceeds to the outer most one. In case of a mismatch the truth index is changed and the mismatching constituents are tagged in order not to produce a mismatch again. Then the comparison process is re-initiated. This results, depending on the form of the sentence and the value of the correct response, in different numbers of constituent comparisons and, hence, in different response latencies. These response latencies are a linear function of the number of constituent comparisons. These constituent comparisons are seen as the basic mental operations which are executed repeatedly. This is indicated by the constant increase in reaction time per added constituent comparison.
Next page

Back to the main page.