Perceptualand Motor Skills, 1991, 72, 299-305.
O Perceptual and Motor Skills 1991
REACTION TIME TO A WORD I N DIFFERENT ORIENTATIONS W I T H LETTERS ROTATED SEPARATELY I N DIFFERENT ORIENTATIONS ' JAMES D. WINDES Northern Arizona Universio Summary.-Six men and six women were timed as they made judgments about a word whose line of letters and individual letters were separately turned in different angular orientations. The subjects were asked to indicate by manual key-pressing responses whether the letters in the word were normal letters or reflected, mirror-image letters. Analysis showed that (1) reaction time (RT) was slower for reflected letters than for normal letters, and for subjects using the left (nondominant) hand rather than the right (dominant) hand for responding to normal letters. (2) RT generally increased in relation to the deviation of the individual letters from normal upright orientation but increased more reliably for normal than for reflected letters. (3) Although there were exceptions, RT for words composed of normal letters generally increased in relation to the deviation of the individual letters from normal upright, regardless of the orientation of the line of letters.
Recent studies (e.g., Koriat & Norman, 1984; Windes & Sjoberg, 1987), have begun to explore the possible relevance of the mental rotation paradigm developed by R. N. Shepard and his associates (see Shepard & Cooper, 1982) to word recognition. Cooper and Shepard (1973) found that RT for distinguishing between normal and reflected letters was longer in relation to their deviation from standard upright orientation, just as Shepard and Metzler (1971) had previously reported for distinguishing between different three-dimensional objects. Windes and Sjoberg required subjects to discriminate between normal and reflected words in different orientations and noted that with normal words, reaction time (RT) increased in relation to the deviation from standard upright orientation in much the same way as Cooper and Shepard found it to do with single letters. With reflected words, however, they failed to find the same type of increase in RT that Cooper and Shepard found with reflected single letters. I n attempting to account for their results with normal words, Windes and Sjoberg conjectured that subjects might have adopted the strategy of searching for and concentrating their attention upon a single letter in the word and then mentally rotating that single letter. This hypothesis, however, could not explain their findings with reflected words. 'This article was presented at the annual meeting of the Rocky Mountain Psychological Association, Tucson, Arizona, A ril 28, 1990. I acknowledge the assistance of Barbara M. Gruber in the data collection a n i of Wayne G . Sjoberg and Angela M. Ellis in anal zing the % ~ ~ , data. Requests for reprints should be sent to ames D Wlndes, Department of ~ s y c h ~ Box 15106, Northern Arizona Univerrity, FlagstaB, Arizona 86011,
300
J. D.WINDES
The general objective of the present study was to explore further the question of what strategy or strategies subjects may use to distinguish normal from reflected letters making up a word. As a means for providing a possible answer to this question, the study employed conditions of orientation of individual letters independently of the orientation of the word in which they occurred. It was hoped that the outcome of this research might have significance for understanding the relationship of letter perception to whole-word recognition in the reading of printed material-something which, despite extensive research over the years, is still poorly understood (Henderson, 1982). Specific research hypotheses were that (1) RT would be slower in relation to the departure of the word (i.e., line of letters) from its normal horizontal position, (2) RT would be slower in relation to the departure of the letters in the word from their normal upright (vertical) position, and (3) RT would be slower for reflected letters than for normal letters.
METHOD Subjects Six men and six women students in introductory psychology were awarded additional course credit for their participation in the experiment. The Edinburgh Handedness Inventory was used to make sure that all subjects were right-handed. Apparatus Stimulus slides were presented on a rear-projection screen by means of a Gerbrands GI176 3-field Projection Tachistoscope, as in the Windes and Sjoberg (1987) study. Manual reaction times (in milliseconds) were recorded on a Hunter Model 120 Klockounter from a Gerbrands Gl36O Reaction Time Apparatus. Stimuli The stimuli were orientational transformations of the word FROG presented either as normal letters or as reflected (mirror-image) letters. The word was presented in block capital letters on Kodalith positive slides. As in the earlier study by Windes and Sjoberg (1987), the entire line of letters in the word appeared in each of six different orientations, indicated in the following by the angle which would be formed by drawing a line perpendicular to the line of letters: 0°, 60°, 120°, 180°, 240°, and 300°. Unlike the earlier study, however, for each of these six different conditions of letter-line or word orientation, the individual letters in the word also appeared in each of the same six possible orientations (0°, 60°, 120°, 180°, 240°, and 300'). For each letter-line orientation, then, there were six different individual letter orientations, making a total of 36 different slides-one for each possible combination of letter-line and individual letter orientations. In six of these
REACTION TIME TO ORIENTATIONS OF WORDS
301
combinations, the word orientation and the individual letter orientations were identical, but for all the rest, the letter-line and letter orientations were different. In one slide, for example, the line of letters was turned at a 60' angle, with each of the letters turned at the same angle. In another, the line of letters was turned at a 60' angle with each of the letters appearing in a normal upright (0°) angle. And so on. For each slide containing normal letters, there was one corresponding slide made up of reflected, mirror-image letters. The reflected-letter slides were made up of letters arranged in a sequence which would correspond to the reflection of the word as a whole. Corresponding to the slide of FROG with both individual letters and line of letters oriented at 0°, for example, there was a slide with reflected individual letters and line of letters also oriented at 0°, and whose letters appeared in the left-to-right sequence of reflected G, reflected 0 , reflected R, and reflected F, since that is the order of such letters when the entire word FROG is reflected as a whole unit. The word appeared at the center of the screen and contained letters which were 2.5 cm high and about 3.8 cm apart from center to center. Since the screen was about 89.0 cm from the subject, words made up of horizontally aligned letters subtended a visual angle of about 1.6O vertically and about 8.7O horizontally. Procedure
Subjects viewed the slides with their heads positioned in a chin-andhead rest and were instructed to fixate upon a central fixation point which appeared whenever the stimulus slides were not exposed. The experimenter signaled the beginning of each exposure trial by closing a switch to sound a tone of 2 sec. duration, after which the slide was exposed for 2 sec. The onset of the stimulus slide in the tachistoscope activated the Klockounter, which was stopped by the subject's manual response. As in one condition of the 1987 study by Windes and Sjoberg, subjects were asked to indicate by their manual key-pressing response whether the letters were normal or reflected letters. Subjects were instructed to press the keys as quickly as possible without making errors. When an error was made, the stimulus was repeated at the end of the series. Half of the subjects (three men and three women) were instructed to press one telegraph key with the right thumb if the stimulus letters were normal and to press a different key with the left thumb if they were reflected letters. The keys were reversed for the other half of the subjects. Thirty-six slides containing normal, nonreflected letters were combined with 36 corresponding slides made up of reflected letters, to make a total of 72 slides that were presented in prearranged random order on three separate days to each of the 12 subjects. The first two days were run for practice, so that only the data collected on the third day were used for statistical analysis.
302
J. D.WINDES
RESULTSAND DISCUSSION The highest error rate was 13.8% for one subject; the next highest was 12.5%. Mean RTs for correct responses were computed for each letter-string under each condition and subjected to analysis of variance. Repeated measures were used across the conditions of letter orientation, line orientation, and letter type (normal or reflected) in a five-way classification factorial, mixed design. The other factors were response hand (left or right) used to indicate normal letters and sex of subject. RT was slower for reflected letters than for normal letters (F,,8= 120.13, p < .01) and for subjects using the left (nondominant) hand to indicate normal letters rather than the right (dominant) hand (Fl,8= 7.38, p < .O5). The effects of letter orientation and line orientation were also significant (F,,,,, = 54.12, p < .01, and F,,,, = 4.43, p < .01, respectively). There was no significant main effect related to sex. 1.10-. ,
-g V1
.oo-
+ Left Hand
+
Rlght Hand
C
0
a c
0.70-
4
0.60-
1'
I
Reflected
Normal
Letler Type
FIG. 1. Mean reaction time as a function of letter type and hand used for response to normal letters
Fig. 1 and the significant interaction between hand and letter type (F,,, = 13.72, p
O Perceptual and Motor Skills 1991
REACTION TIME TO A WORD I N DIFFERENT ORIENTATIONS W I T H LETTERS ROTATED SEPARATELY I N DIFFERENT ORIENTATIONS ' JAMES D. WINDES Northern Arizona Universio Summary.-Six men and six women were timed as they made judgments about a word whose line of letters and individual letters were separately turned in different angular orientations. The subjects were asked to indicate by manual key-pressing responses whether the letters in the word were normal letters or reflected, mirror-image letters. Analysis showed that (1) reaction time (RT) was slower for reflected letters than for normal letters, and for subjects using the left (nondominant) hand rather than the right (dominant) hand for responding to normal letters. (2) RT generally increased in relation to the deviation of the individual letters from normal upright orientation but increased more reliably for normal than for reflected letters. (3) Although there were exceptions, RT for words composed of normal letters generally increased in relation to the deviation of the individual letters from normal upright, regardless of the orientation of the line of letters.
Recent studies (e.g., Koriat & Norman, 1984; Windes & Sjoberg, 1987), have begun to explore the possible relevance of the mental rotation paradigm developed by R. N. Shepard and his associates (see Shepard & Cooper, 1982) to word recognition. Cooper and Shepard (1973) found that RT for distinguishing between normal and reflected letters was longer in relation to their deviation from standard upright orientation, just as Shepard and Metzler (1971) had previously reported for distinguishing between different three-dimensional objects. Windes and Sjoberg required subjects to discriminate between normal and reflected words in different orientations and noted that with normal words, reaction time (RT) increased in relation to the deviation from standard upright orientation in much the same way as Cooper and Shepard found it to do with single letters. With reflected words, however, they failed to find the same type of increase in RT that Cooper and Shepard found with reflected single letters. I n attempting to account for their results with normal words, Windes and Sjoberg conjectured that subjects might have adopted the strategy of searching for and concentrating their attention upon a single letter in the word and then mentally rotating that single letter. This hypothesis, however, could not explain their findings with reflected words. 'This article was presented at the annual meeting of the Rocky Mountain Psychological Association, Tucson, Arizona, A ril 28, 1990. I acknowledge the assistance of Barbara M. Gruber in the data collection a n i of Wayne G . Sjoberg and Angela M. Ellis in anal zing the % ~ ~ , data. Requests for reprints should be sent to ames D Wlndes, Department of ~ s y c h ~ Box 15106, Northern Arizona Univerrity, FlagstaB, Arizona 86011,
300
J. D.WINDES
The general objective of the present study was to explore further the question of what strategy or strategies subjects may use to distinguish normal from reflected letters making up a word. As a means for providing a possible answer to this question, the study employed conditions of orientation of individual letters independently of the orientation of the word in which they occurred. It was hoped that the outcome of this research might have significance for understanding the relationship of letter perception to whole-word recognition in the reading of printed material-something which, despite extensive research over the years, is still poorly understood (Henderson, 1982). Specific research hypotheses were that (1) RT would be slower in relation to the departure of the word (i.e., line of letters) from its normal horizontal position, (2) RT would be slower in relation to the departure of the letters in the word from their normal upright (vertical) position, and (3) RT would be slower for reflected letters than for normal letters.
METHOD Subjects Six men and six women students in introductory psychology were awarded additional course credit for their participation in the experiment. The Edinburgh Handedness Inventory was used to make sure that all subjects were right-handed. Apparatus Stimulus slides were presented on a rear-projection screen by means of a Gerbrands GI176 3-field Projection Tachistoscope, as in the Windes and Sjoberg (1987) study. Manual reaction times (in milliseconds) were recorded on a Hunter Model 120 Klockounter from a Gerbrands Gl36O Reaction Time Apparatus. Stimuli The stimuli were orientational transformations of the word FROG presented either as normal letters or as reflected (mirror-image) letters. The word was presented in block capital letters on Kodalith positive slides. As in the earlier study by Windes and Sjoberg (1987), the entire line of letters in the word appeared in each of six different orientations, indicated in the following by the angle which would be formed by drawing a line perpendicular to the line of letters: 0°, 60°, 120°, 180°, 240°, and 300°. Unlike the earlier study, however, for each of these six different conditions of letter-line or word orientation, the individual letters in the word also appeared in each of the same six possible orientations (0°, 60°, 120°, 180°, 240°, and 300'). For each letter-line orientation, then, there were six different individual letter orientations, making a total of 36 different slides-one for each possible combination of letter-line and individual letter orientations. In six of these
REACTION TIME TO ORIENTATIONS OF WORDS
301
combinations, the word orientation and the individual letter orientations were identical, but for all the rest, the letter-line and letter orientations were different. In one slide, for example, the line of letters was turned at a 60' angle, with each of the letters turned at the same angle. In another, the line of letters was turned at a 60' angle with each of the letters appearing in a normal upright (0°) angle. And so on. For each slide containing normal letters, there was one corresponding slide made up of reflected, mirror-image letters. The reflected-letter slides were made up of letters arranged in a sequence which would correspond to the reflection of the word as a whole. Corresponding to the slide of FROG with both individual letters and line of letters oriented at 0°, for example, there was a slide with reflected individual letters and line of letters also oriented at 0°, and whose letters appeared in the left-to-right sequence of reflected G, reflected 0 , reflected R, and reflected F, since that is the order of such letters when the entire word FROG is reflected as a whole unit. The word appeared at the center of the screen and contained letters which were 2.5 cm high and about 3.8 cm apart from center to center. Since the screen was about 89.0 cm from the subject, words made up of horizontally aligned letters subtended a visual angle of about 1.6O vertically and about 8.7O horizontally. Procedure
Subjects viewed the slides with their heads positioned in a chin-andhead rest and were instructed to fixate upon a central fixation point which appeared whenever the stimulus slides were not exposed. The experimenter signaled the beginning of each exposure trial by closing a switch to sound a tone of 2 sec. duration, after which the slide was exposed for 2 sec. The onset of the stimulus slide in the tachistoscope activated the Klockounter, which was stopped by the subject's manual response. As in one condition of the 1987 study by Windes and Sjoberg, subjects were asked to indicate by their manual key-pressing response whether the letters were normal or reflected letters. Subjects were instructed to press the keys as quickly as possible without making errors. When an error was made, the stimulus was repeated at the end of the series. Half of the subjects (three men and three women) were instructed to press one telegraph key with the right thumb if the stimulus letters were normal and to press a different key with the left thumb if they were reflected letters. The keys were reversed for the other half of the subjects. Thirty-six slides containing normal, nonreflected letters were combined with 36 corresponding slides made up of reflected letters, to make a total of 72 slides that were presented in prearranged random order on three separate days to each of the 12 subjects. The first two days were run for practice, so that only the data collected on the third day were used for statistical analysis.
302
J. D.WINDES
RESULTSAND DISCUSSION The highest error rate was 13.8% for one subject; the next highest was 12.5%. Mean RTs for correct responses were computed for each letter-string under each condition and subjected to analysis of variance. Repeated measures were used across the conditions of letter orientation, line orientation, and letter type (normal or reflected) in a five-way classification factorial, mixed design. The other factors were response hand (left or right) used to indicate normal letters and sex of subject. RT was slower for reflected letters than for normal letters (F,,8= 120.13, p < .01) and for subjects using the left (nondominant) hand to indicate normal letters rather than the right (dominant) hand (Fl,8= 7.38, p < .O5). The effects of letter orientation and line orientation were also significant (F,,,,, = 54.12, p < .01, and F,,,, = 4.43, p < .01, respectively). There was no significant main effect related to sex. 1.10-. ,
-g V1
.oo-
+ Left Hand
+
Rlght Hand
C
0
a c
0.70-
4
0.60-
1'
I
Reflected
Normal
Letler Type
FIG. 1. Mean reaction time as a function of letter type and hand used for response to normal letters
Fig. 1 and the significant interaction between hand and letter type (F,,, = 13.72, p
