Abstract
During proofreading, some anomalous words can be detected easily and others with difficulty. The characteristics of proofreading errors may contribute to understanding some part of the reading process. Reading is a complicated cognitive process and depends on so many factors. Assessing the degree of influence of each factor gives us insight into the reading mechanisms. Using multiple regression analysis, we assessed the contribution of factors that each anomalous word had (e.g., phonological or semantic similarity to the contextually correct word, location in sentence, and size of the span of text needed to detect the anomalous word) to the effectiveness of anomalous word detection in proofreading. Novices and professionals in proofreading participated in the experiment. The results showed that the size of the span of text needed for anomalous word detection had especially large influence on anomalous word detection, and that professional proofreaders have special characteristics in processing span of text.