In most experiments dealing with the relative effectiveness of different word parts in word identification, the greater importance of the word beginning has been ascribed to the sequential order of speech. However, differences in the a m o u n t of information must also be taken into account: initial letters contain more information than final letters. In order to determine whether both factors have an effect, an experiment was carried out in which 48 Ss had to guess Dutch 7-1etter nouns from a varying number of letters which constituted either the initial or the final word part. For these nouns as a group, beginnings and endings carried equal amounts of information.The results indicated that both information and serial order in speech were effective. The time required for identification was dependent on the amount of information of the n-gram presented. The Ss also enumerated more 7-letter nouns if the initial letters wer e available, and as a result identification took less time. In addition, the enumerated nouns were found to be re!atively f r e q u e n t words, and speed of solution was directly related to frequency of occurrence in the language.Tachistoscopically presented words yield better recognition in the field at the right side of the fixation point than in the left peripheral field, as is shown in several experiments in which the recognition threshold for words was investigated (Mishkin and Forgays, 1952;Forgays, 1953). Melville (1957) showed that this difference in recognition time is larger for 7-letter words than for 3-letter words. In these experiments, the beginning of the words on the right side of the fixation point is more central in the field of vision than the initial letters of the words on the left side, and the longer the words, the larger this difference. These results indicate the relatively larger importance of the beginning of the word. The same conclusion was reached by Bruner and O'Dowd (1958) who showed that tachistoscopic recognition of words was faster when the final letters were reversed than when the initial letters were reversed.This difference might have two reasons: (a) habits in speech, favoring the beginning as a starting point for producing and recognizing words, and (b) the differential amount of information in initial and final letters. Miller and Friedman (1957), in their explanation of the same asymmetrical effect for text passages, stress the former point. In word recognition, however, the amount of information has to be considered as well (Aborn and Rubenstein, 1960). The information, depending on the number of alternatives and the probabilities of these alternatives, can be measured in two ways: (a) the number of different words in which an n-gram (i.e., n sequential letters) occurs, and (b) the relative frequency of occurrence of the n-gram in the language. The correlation between these two measures is high, and both can be used as information indicators for the n-grams in relation to words.In the English language the n-grams at 441
