Journal of Experimental Psychology. Human Perception and Performance 1989. Vol. 15. No. I, 142-152 Copyright 1989 by the American Psychological Association. Inc. 00%-1523/89/500.75 Reading Long Words Embedded in Sentences: Informativeness of Word Halves Affects Eye Movements Jukka Hyona and Pekka Niemi University of Turku, Turku, Finland Geoffrey Underwood University of Nottingham, Nottingham, England The possibility was explored that the informativeness of a specific region within a word can influence eye movements during reading. In Experiment 1, words containing identifying infor- mation either toward the beginning or toward the end were displayed asymmetrically around the point of fixation so that the reader was initially presented with either the informative or noninformative zone. Words were read with shorter summed initial fixation time when the reading was started from the informative zone. In Experiments 2 and 3, the target words were presented in sentences that were to be comprehended. More attention was given to the informative endings of words than to redundant endings. The latter were also skipped more often. The duration of the first fixation was not affected by information distribution within the word, whereas the second fixation duration was. The results of these experiments lend good support to the hypothesis of immediate lexical control over fixation behavior and to the notion of a convenient viewing position. Three experiments deal with two related issues in the in- vestigation of eye guidance during reading. First, does the information distribution within a word influence fixation time on that word? Second, are fixation locations affected by the information distribution within words? These questions were investigated by observing the eye fixations of readers who were presented with words containing uneven distributions of information. The analyses are based on fixations in two different locations: those coinciding with the informative part of a target word and those coinciding with the uninformative part. There is evidence that the information value of a word may affect the pattern of eye movements during reading. It has been demonstrated that syntactic constraints can affect fixa- tion patterns. Carpenter and Just (1983) and O'Regan (1979) reported that subjects fixated three-letter function words less frequently than content words of equal length. An effect of semantic constraint was observed by Ehrlich and Rayner (1981). Specific words in predictable contexts were fixated less often than the same words in neutral contexts, and when A preliminary report of these experiments was made at the Third European Conference on Eye Movements, Dourdan, France, Septem- ber 1985. This work was supported by grants from the Research Council for the Social Sciences of the Academy of Finland to Pekka Niemi, and Project Grant G8602578N from the Medical Research Council (UK.) to Geoffrey Underwood. We wish to thank Juhani Tuominen for consultations during the data analysis, and Thomas Carr, James Cutting, William Epstein, Kevin O'Regan, Keith Rayner, and an anonymous referee for con- structive criticisms of earlier versions of the article. Correspondence concerning this article should be addressed to Pekka Niemi, who is now at the Department of Psychology, Abo Akademi (The Swedish University of Turku), Nunneg. 4, SF-20700 Turku, Finland. they were fixated, they received shorter fixations. These are examples of how words that carry more information are fixated more often and longer than more redundant words. It is also possible that different parts of a single word may carry different amounts of information. Could this be reflected in readers' fixation patterns? This question was first posed by O'Regan, I.evy-Schoen, Pynte, and Brugaillere (1984, Exper- iment 3) and O'Regan (1984), who used target words long enough to attract more than one fixation. The distinctive feature of these words was that information vital for identi- fying the word was located either within the first six letters or within the final six letters. The readers were asked to judge whether two words were semantically related. The target words were presented asymmetrically about the point of fix- ation, with fixation coinciding either with the third letter from the beginning or the third letter from the end of the word. The idea is that if fixations are under the moment-to-moment control of ongoing processing, then being forced to fixate a noninformative part of a word makes the reader move quickly away from that part. Therefore, the informative part of the same word would receive longer fixations. The findings of O'Regan and his associates lent partial support to this hy- pothesis. The information-at-the-end [info(end)] words were read with relatively short gaze durations when reading started from the end (i.e., from the informative zone). Gaze durations were slightly longer when reading started from the beginning of the word. Analogously, the information-at-the-beginning [info(beg)] words received short gaze durations when reading started from the beginning of the word, and longer gaze durations when the initial point of fixation coincided with the less informative end zone. O'Regan (1984) performed a separate analysis concerning eye movement patterns involving either one or two fixations on a word. When reading started from the beginning, 22% of the info(beg) words were read with a single fixation, whereas 142