Ординатура / Офтальмология / Английские материалы / Eye Movements A Window on Mind and Brain_Van Gompel_2007

when only fixations that are launched from a near position were considered, the 1st constituent length did not modulate the display change effect. Other evidence against the morphological preprocessing hypothesis comes from the duration of fixation prior to fixating the target and from the first fixation location; neither of them was affected by the length of the 1st constituent or by the type of parafoveal preview. That 1st constituent length did not influence first fixation location compares favourably with the results of Experiment 1 of Hyönä and Pollatsek (1998).

Effects of 1st constituent length on foveal processing replicated the results of Hyönä and Pollatsek (1998) also in other respects. First, the second fixation location was further into the word when the 1st constituent was long. This lengthening of the first within-word saccade is needed when attempting to optimize the fixation location on the 2nd constituent. The constituent length effect in gaze duration was significant in the present study (an effect of 33 ms), but only marginal in Hyönä and Pollatsek (an effect of 10 ms). The trade-off in the duration of first versus second fixation (a longer first fixation but a shorter second fixation for the short 1st constituent compounds) was also observed in Hyönä and Pollatsek. In addition, there was a hint for the third fixation duration to be shorter for short 1st constituent compounds. Another indication that long 1st constituent compounds are processed differently from short 1st constituent compounds came from the regression analyses. These analyses showed that 1st constituent frequency but not 2nd constituent frequency was a reliable predictor of processing long 1st constituent compounds, whereas 2nd constituent frequency, but not 1st constituent frequency, predicted gaze duration on short 1st constituent compounds. For both types of compounds the whole word frequency turned out to be a reliable secondary predictor.

Finally, the finding that display change increased the number of fixations (the effect was not quite significant in the item analysis) may be taken to suggest that more preprocessing was done in the full preview condition, thus diminishing slightly the need for making additional refixations on the target word.

8. Discussion

In contrast to the prediction of Kambe (2004), a morphological preview benefit in Finnish seems hard to obtain. Thus, even though Finnish is a morphologically rich and highly productive language, readers do not seem to make use of parafoveally available morphological codes. Extracting a morphological unit out of a multimorphemic word is not an easy task, not even during foveal processing. As mentioned in the Introduction, Bertram et al. (2004) found evidence that morphological parsing of long 1st constituent compound words without a clear morpheme boundary cue is a time-consuming process. This is most probably so because foveal vision is needed to locate a morpheme boundary that is not clearly marked. When the to-be-located morpheme boundary is in the parafoveally presented word, the task may become practically impossible. Apart from that, assuming that morphological codes become activated after orthographic codes, there simply may not be enough time for a morphological unit to accumulate enough activation when it is

available in the parafovea. Evidence in support of this hypothesis comes from Kambe’s (2004) second experiment, in which she failed to find a morphological preview benefit, even though she demarcated morpheme boundaries in a highly salient way by using capital Xs (e.g., reXXXX). Even when she considered only fixations launched from a near position, the morphological preview benefit failed to show up. This implies that, at least in English, a morphological preview benefit may be impossible to obtain. The current study indicates also that, in Finnish, parafoveal processing is restricted to the visual-orthographic level. The fact that the preview effect, which appeared mainly in gaze duration and in the number of first-pass fixations, was small (though not smaller in size than in other studies using a similar manipulation) implies that the first 3–4 letters are of greatest importance in the word identification process (as also argued by Briihl & Inhoff, 1995). However, the finding that a change effect is nevertheless observable when the first 3–4 letters were preserved while the remaining letters were replaced by random letters entails that also in Finnish non-initial letters are coded in the parafovea to some extent.

In line with previous studies, the present study showed that once a compound is fixated, morphological structure is recognized and affects processing. In addition, constituent length and the length of the whole compound seem to play a crucial role in how the compound word is processed. We observed that compounds of equal length, but with different 1st and thus 2nd constituent lengths, elicited differences in gaze duration, as well as in first, second and third fixation duration, and second fixation location. A short 1st constituent compound elicited longer first fixation durations but shorter second, third, and gaze durations than a long 1st constituent compound. In addition, for short 1st constituent compounds the second fixation was located around 0.8 characters closer to the word beginning and gaze duration was mainly determined by 2nd constituent frequency, whereas for long 1st constituent compounds gaze duration varied primarily as a function of 1st constituent frequency. This pattern of results implies that a reader can deal more easily with a long compound, when it has a short 1st constituent. Perhaps orthographic preprocessing of the initial trigram in addition to the whole 1st constituent being in clear foveal vision once the compound word is fixated makes the 1st constituent readily accessible. This may even lead to a situation where a 1st constituent frequency effect, normally observed for long Finnish compounds, is undermined. The results of Juhasz, Starr, Inhoff, & Placke, (2003) are compatible with this line of reasoning. Juhasz et al. found a much more prominent role for 2nd constituent frequency than for 1st constituent frequency in English compounds with a 1st constituent of 3–5 characters (average 4.6). Their claim is that a compound may be accessed via the 2nd constituent, since that is the main carrier of compound meaning. However, the question is then how to morphologically decompose a compound so that lexical-semantic information of the 1st constituent is not accessed or is even totally neglected. It seems more plausible to assume that due to visual acuity reasons a short 1st constituent is more readily available than a long 1st constituent, and that this allows fast lexical access to the first constituent, even when 1st constituent frequency is not so high. Thus it seems that visual acuity constraints and a word’s morphological structure interact in interesting ways during the foveal processing of long Finnish compound words. However, to obtain a more detailed insight into the interaction

of 1st constituent length and compound processing, more systematic experimentation is needed.

The most important conclusion of this chapter is nevertheless that despite the morphological productivity of the Finnish language, readers of Finnish do not seem to profit from a parafoveal morphological preview. Hence, readers of Hebrew remain unique in being able to crack the morphological code parafoveally.

Ackowledgements

This study was financially supported by the Academy of Finland (grant to the second author).

References

Balota, D. A., Pollatsek, A., & Rayner, K. (1985). The interaction of contextual constraints and parafoveal visual information in reading. Cognitive Psychology, 17, 364–390.

Bertram, R., & Hyönä, J. (2003). The length of a complex word modifies the role of morphological structure: Evidence from eye movements when reading short and long Finnish compounds. Journal of Memory and Language, 48, 615–634.

Bertram, R., Pollatsek, A., & Hyönä, J. (2004). Morphological parsing and the use of segmentation cues in reading Finnish compounds. Journal of Memory and Language, 51, 325–345.

Briihl, D., & Inhoff, A. W. (1995). Integrating information across fixations during reading: The use of orthographic bodies and exterior letters. Journal of Experimental Psychology: Learning, Memory & Cognition, 21, 55–67.

Deutsch, A., Frost, R., Pelleg, S., Pollatsek, A., & Rayner, K. (2003). Early morphological effects in reading: Evidence from parafoveal preview benefit in Hebrew. Psychonomic Bulletin & Review, 10, 415–422.

Deutsch, A., Frost, R., Pollatsek, A., & Rayner, K. (2000). Early morphological effects in word recognition in Hebrew: Evidence from parafoveal preview benefit. Language and Cognitive Processes, 15, 487–506.

Deutsch, A., Frost, R., Pollatsek, A., & Rayner, K. (2005). Morphological parafoveal preview benefit effects in reading: Evidence in Hebrew: Evidence from parafoveal. Language and Cognitive Processes, 20, 341–371.

Hyönä, J., & Bertram, R. (2004). Do frequency characteristics of nonfixated words influence the processing of fixated words during reading? European Journal of Cognitive Psychology, 16, 104–127.

Hyönä. J., Bertram, R., & Pollatsek, A. (2004). Are long compounds identified serially via their constituents? Evidence from an eye-movement contingent display change study. Memory and Cognition, 32, 523–532.

Hyönä. J., Bertram, R., & Pollatsek, A. (2005). Identifying compound words in reading: An overview and a model. In G. Underwood (Ed.), Cognitive processes in eye guidance (pp. 79–104). Oxford: University Press.

Hyönä, J., & Pollatsek, A. (1998). Reading Finnish compound words: Eye fixations are affected by component morphemes. Journal of Experimental Psychology: Human Perception & Performance, 24, 1612–1627.

Inhoff, A. W. (1989). Parafoveal processing of words and saccade computation during eye fixations in reading.

Journal of Experimental Psychology:Human Perception and Performance, 15, 544–555.

Juhasz, B. J., Starr, M. S., Inhoff, A. W., & Placke, L. (2003). The effects of morphology on the processing of compound words: Evidence from naming, lexical decisions and eye fixations. British Journal of Psychology, 94, 223–244.

Kambe, G. (2004). Parafoveal processing of prefixed words during eye fixations in reading: Evidence against morphological influences on parafoveal processing. Perception & Psychophysics, 66, 279–292.

Kennedy, A. (2000). Parafoveal processing in word recognition. Quarterly Journal of Experimental Psychology, 53A, 429–455.

Laine, M., & Virtanen, P. (1999). WordMill lexical search program. Center for Cognitive Neuroscience, University of Turku, Finland.

Lima, S. D., (1987). Morphological analysis in sentence reading. Journal of Memory and Language, 26, 84–99. Lorch, R. F., & Myers, J. L. (1990). Regression analyses of repeated measures data in cognitive research.

Journal of Experimental Psychology; Learning, Memory, and Cognition, 16, 149–157.

Murray, W. S. (1998). Parafoveal pragmatics. In G. Underwood (Ed.), Eye guidance in reading and scene perception (pp. 181–199). Oxford: Elsevier Science.

O’Regan, J. K. (1992). Optimal viewing position in words and the strategy-tactics theory of eye movements in reading. In K. Rayner (Ed.), Eye movements and visual cognition: scene perception and reading

(pp. 334–354). New York: Springer-Verlag.

Pollatsek, A., Rayner, K., & Balota, D. A. (1986). Inferences about eye movement control from the perceptual span in reading. Perception & Psychophysics, 40, 123–130.

Rayner, K. (1975). The perceptual span and peripheral cues in reading. Cognitive Psychology, 7, 65–81. Rayner, K. (1979). Eye-guidance in reading: Fixation location within words. Perception, 8, 21–30.

Rayner, K., Balota, D. A., & Pollarsek, A. (1986). Against parafoveal semantic preprocessing during eye fixations in reading. Canadian Journal of Psychology, 40, 473–483.

Rayner, K., White, S. J., Kambe, G., Miller, B., & Liversedge, S. P. (2003). On the processing of meaning from parafoveal vision during eye fixations in reading. In J. Hyönä, R. Radach, & H. Deubel (Eds.), The mind’s eye: cognitive and applied aspects of eye movement research (pp. 213–234). Oxford: Elsevier Science.

This page intentionally left blank

Abstract

Three experiments showed that localised foveal load does not modulate the probability of skipping the following 4–6 letter parafoveal word (word n + 1). In Experiments 1 and 2 the preview of the word n + 1 was always correct. In Experiment 3 the preview of the word n+1 was either correct or incorrect. Localised foveal difficulty did not significantly modulate the effect of preview on the probability of skipping the word n + 1. The results suggest that the processes that produce modulations of parafoveal preprocessing by foveal load on reading time measures may not apply to the control of word skipping.

n + 1 (whilst fixating word n) is restricted by the time required to complete processing of word n. Critically, the time to attend to word n + 1 influences the extent to which word n + 1 is preprocessed and therefore the amount of preview benefit for word n + 1 (as shown by reading times on word n+1). Similarly, once attention has moved to word n+1, if there is sufficient time before the saccade is executed, and if word n + 1 is identified quickly enough, then the saccade programme may be re-programmed to skip word n + 1. Therefore both reading times and word skipping are determined by the same mechanism which is influenced by foveal load (time to process word n). Consequently Reichle et al. predict that foveal load should reduce the probability of skipping the following word.

In contrast, in their Glenmore model, Reilly and Radach (2003) suggest that multiple words can be processed in parallel and that there is competition between words for activation related to linguistic processing. Consequently greater processing of the fixated word reduces processing of other words. As a result, foveal load reduces preview benefit for the following word. Reilly and Radach also suggest that each word has a salience value, such that saccades are directed to the word with greatest salience. These salience values are influenced by the same linguistic activation system that influences reading times. Therefore, although not explicitly stated, Reilly and Radach’s account also appears to suggest that word skipping is modulated by foveal load.

To summarise, empirical evidence suggests that foveal load modulates parafoveal preprocessing as shown by reading time preview benefit. Accounts based on both serial processing and parallel processing have been proposed that explain this phenomenon. Both these models also predict that foveal load modulates word skipping in a similar way as for preview benefits. However, other studies have suggested that the processes that determine when and where the eyes move can be different (Radach & Heller, 2000; Rayner & McConkie, 1976; Rayner & Pollatsek, 1981). Indeed, a number of models of eye movement control in reading have been developed in which different mechanisms determine when and where the eyes move. These models either do not predict that foveal load modulates preprocessing as in the case of SWIFT (Engbert, Longtin, & Kliegl, 2002; Engbert, Nuthmann, Richter & Kliegl, 2005; Kliegl & Engbert, 2003) and the Competition/Interaction model (Yang & McConkie, 2001), or they predict that both the when and the where systems are modulated by foveal load, as in Glenmore (Reilly & Radach, 2003). Nevertheless, accounts which differentiate between mechanisms that determine when and where the eyes move highlight the possibility that although foveal load modulates preprocessing as shown by reading times, foveal load may not necessarily modulate where the eyes move (as shown by the probability of word skipping).

The issue of whether foveal load modulates the probability of word skipping is therefore critical not only for evaluating the architecture of current models but also for assessing the fundamental question of whether the mechanisms that determine when and where the eyes move are the same or different. Drieghe et al. (2005) investigated whether foveal load modulated the probability of skipping parafoveal three-letter words (see also Kennison & Clifton, 1995). For cases in which there was a correct preview of the parafoveal word, Drieghe et al. showed no significant effect of foveal load on the probability of skipping the following word. Despite the non-significant result, skipping rates were numerically

higher when there was low, compared to high, foveal load which is suggestive of the possibility that foveal load may modulate word skipping. Therefore it is important to examine whether foveal load does reliably influence word skipping. Also, as Drieghe et al. only tested the probability of skipping three letter words, it is important to test whether foveal load modulates the probability of skipping slightly longer words.

The present study includes three experiments that test whether localised foveal load influences the probability of skipping fourto six-letter parafoveal words. The manipulations of foveal load include orthographic regularity, spelling and word frequency. These manipulations are intended to influence the ease with which a specific word can be processed. Importantly, this study does not test whether general processing load modulates word skipping. General processing load may be modified by text difficulty (e.g. contextual factors) or reading strategy. For example, general processing load could modulate global parameters for eye movement control such that increased load might increase fixation durations or shorten saccade lengths (see Yang & McConkie, 2001). Within each of the experiments presented here, such general factors are controlled by using the same sentence beginnings up until the critical words across each of the experimental conditions.

For all of the experiments presented here, the analyses include only cases in which a single fixation was made on the foveal word and no regressions were made out of the foveal word. Refixations on the foveal word could modify factors which might influence word skipping, such as launch site and the quality of the parafoveal preview. Therefore, restricting the analyses to cases in which single fixations were made on the foveal word ensures that any differences in skipping probabilities could not be accounted for by differences in refixation probabilities. Overall, if foveal load influences word skipping then the probability of skipping the parafoveal word (word n + 1) should be greater when the foveal word (word n) is easy, compared to difficult, to process.

1. Experiment 1

In Experiment 1, foveal load was manipulated by orthographic regularity. The foveal word (word n) was either orthographically regular (low foveal load e.g. miniature) or orthographically irregular (high foveal load e.g. ergonomic) and was followed by the parafoveal word (word n + 1) (e.g. chairs).

Note that the foveal processing load manipulation in Experiment 1 has been shown to have a small (less than 0.5 character) but reliable influence on initial fixation positions on these words (White & Liversedge, 2006a). Therefore in the present study, initial fixations land nearer to the beginning of the foveal words that are difficult to process (orthographically irregular) than the foveal words that are easy to process (orthographically regular). Consequently, the launch site prior to skipping or fixating word n − 1 may have been slightly further away for the high foveal load words compared to the low foveal load words. Launch site may influence skipping probabilities such that saccades launched from further away may be less likely to skip word n + 1. Importantly, note that the direction of these effects would have facilitated an effect of foveal load on word skipping such that