Word Recognition

Word Recognition as a Perceptual Choice

People recognize the words of their language remarkably quickly and accurately, whether written or spoken. [Page 1142]However, this ease conceals the complexity of the process. The roughly 50,000 words that an adult knows are formed by combining sound or letter elements drawn from a much smaller set (e.g., 26 letters and between 40 and 45 distinct sounds in English, depending on the dialect). As a result, words resemble one another to a large degree. The physical difference between the words cat and cot is subtle but critical to attend to if the correct meaning is to be retrieved. Thus, recognition entails selecting the best matching hypothesis among alternatives.

Importantly, these alternatives exert an influence on the outcome. Under some conditions, words are harder to recognize if they have many “neighbors,” that is, if they resemble many other words. For instance, while reading, people spend more time fixating on printed words with many neighbors than words with few, suggesting that the former take longer to identify. Likewise, spoken words that sound similar to other words are recognized more slowly, and less accurately when presented in noise, than spoken words with fewer neighbors. These findings are often taken as indicating that recognition is a competitive process, where the spoken or written stimulus is compared to every hypothesis simultaneously and where the best matching hypothesis is identified with respect to the overall support accrued by the alternatives. However, under some conditions, similarity to many other words facilitates a word's recognition. For example, written words that share all but one of their letters with many other words (e.g., cave, similar to at least 16 other words such as cake, have, gave, care) are recognized faster than words with fewer such neighbors (e.g., next, similar to only 4 other words, newt, neat, nest, and text). That neighbors can facilitate and impede a word's recognition has been taken as reflecting the existence of two separate mechanisms by which neighbors affect a word's recognition. Neighbors provide mutually incompatible interpretations of the same portion of the stimulus, the part that differs among them. The more alternatives, the harder the perceptual choice becomes. But neighbors also share some elements, and the perception of these parts is enhanced by the support provided by their occurrence in multiple words. Whether neighbors facilitate or impede the overall outcome depends on which influence outweighs the other.

...