Neuropsychological studies of reading and writing
This thesis investigates the reading and writing of two patients with brain injuries due to cerebro-vascular accidents. Background tests show both patients to be moderately anomic and to have severe impairments in reading and writing nonwords. Investigations of the locus of impairment in AN's nonword reading showed her to have normal orthographic analysis capabilities but impairments in converting single and multiple graphemes into phonemes and in phonemic blending. The central issue studied was the role of lexical but non-semantic processes in reading aloud, writing to dictation and copying. For this purpose a "familiar nonword" paradigm was developed in which the patients learned to read or write a small set of nonwords either with or without any associated semantics. Both AN and AM were able to learn to read nonwords to which no meanings were attached but they could still not read novel nonwords. Both patients were unable to report any meanings for the familiar nonwords when they read them and there was no evidence that learning to read them improved their sub-lexical processing abilities. These results are evidence for a direct lexical route from print to sound that is dedicated to processing whole familiar words. It was also shown with AN that if nonwords are given meanings then learning is faster than if they are not given meanings. Experiments designed to test the hypothesis that nonwords are read by analogy to words found no support for it. Both patients have severe impairments in writing novel nonwords to dictation. As they can repeat spoken nonwords after they have failed to write them, this is not due to a short-term memory impairment. Despite their nonword writing impairments, both patients were able to write to dictation the meaningless nonwords that they had previously learned to read at the first attempt, and AN did so one month after learning to read them. Neither patient however, could write novel nonwords made by reordering the letters of the familiar nonwords. Furthermore, the familiar nonwords used spellings that are of a priori low probability. The familiar nonwords must therefore have been written using lexical knowledge. Tests of semantic association showed that the familiar nonwords evoked no semantic information that the patients could report. Function words dictated to AN evoked little semantic information but she wrote them to dictation significantly better than nonwords made by reordering their letters. These results are evidence for a direct lexical route for writing to dictation. Copying was studied both with and without a five second delay between presentation and response. AN was better at delayed copying of meaningless but familiar nonwords than she was at copying novel nonwords. She was also better at delayed copying of six-letter, bi-syllabic nonwords that she had been trained to copy than she was at copying novel nonwords made by recombining the first and second halves of the familiar nonwords such that these halves retained their positions from the parent nonwords. AN was better at copying function words than nonwords made by reordering their letters. She was also better at copying function words than she was at reading or writing them to dictation. These results are evidence for a direct lexical route for copying. AN and AM were both able to write to dictation nonwords that they had never heard or written before but with which they had been made visually familiar during a visual discrimination task. They must have used lexical knowledge to do so because the spellings used were of a priori very low probability. The creation of lexical orthographic information which can be retrieved from novel auditory input raises difficulties for current models and various possible interpretations are discussed. Finally, some of the possible implications of the re-learning abilities shown by these patients, for rehabilitation procedures are discussed briefly.