Perceptnal and Motor Skills, 1979, 49, 67-70. @ Perceptual and Motor Skills 1979,
PICTURE-NAMING INTERFERENCE W I T H GOOD AND POOR READERS1 ROBERT E. GUTTENTAG University of Denver Strmmary.-22
third grade children of high and low reading ability (ns
= 1 1 ) were tested for their ability to name pictures while trying to ignore words or nonword strings of letters printed inside the pictures. Nonoverlapping sets of pictures and words were used as stimuli to avoid the possibility of sensitizing subjects to the particular words used in the experiment (Neisser, 1976). Both groups experienced more interference from intra-category than extracategory words, indicating that they processed the words automatically. Only the good readers experienced more interference from pseudowords than consonant strings, suggesting that poor readers are less sensitive than good readers to orthographic regularity.
In 1935 Stroop devised a task requiring subjects to name the ink colors of printed color-words while trying to ignore the words themselves. ,Skilled readers have considerable difficulty ignoring the words (Dyer, 1973), suggesting that they process isolated printed word automatically. Several recent studies employing a picture-word variant of the Stroop task have demonstrated chat even children who are poor readers experience interference from to-beignored printed words, suggesting that poor readers are not deficient at processing printed words automatically (Golinkoff & Rosinski,' 1976; Guttentag & Haith, 1978). Neisser ( 1976), however, has questioned the generalizability of findings from the Stroop task on the grounds that the use of interfering words which correspond to task-relevant responses may sensitize subjects to the specific words used in the experiment. This criticism applies as well to the studies which have used pictures and words as stimuli. The present study examined automatic word processing by good and poor readers with a procedure designed to avoid sensitizing subjects to the particular words used in the experiment. This was accomplished by using nonoverlapping sets of pictures and words as stimuli. For example, one of the distractor words used in the present study was "bird," but at no time were subjects required to name a picture of a bird. Hence, there was no reason for subjects to be sensitized to process the printed word "bird." T o determine whether subjects extracted meaning automatically from the printed words, the categorical relation between each distractor word and the picture it was princed 'This research was supported by a doctoral fellowship from the Canada Council to the author and by a grant from NICHHD ( H D 0 7 2 3 2 ) to Marshall M. Haith. The research would not have been possible without the cooperation of Charles Stanley, Principal, and the staff and students of Cherry Hills Elementarv School. Reprint requests should ~, o'f Winnipeg, be sent to Robert E. Guttentag, Department of ~ s ; c h o l o ~LJni;ersity Winnipeg, Manitoba R3B 2E9.
within was systematically varied. If subjects experience more interference from intra-category words, e.g., a picture of a dog containing the word "fish," than from extra-category words, e.g., a picture of a dog containing the word "lamp," then we may conclude that they have extracted meaning from the printed words. And since their attention is directed at picture naming rather than word processing, we may condude that they have processed the words automatically. The present study also examined the amount of picture-naming interference produced by nonword strings of letters. An earlier study by Guttentag and Haith (1978) suggested that only good readers experience more interference from pseudowords than from consonant strings. If this finding, which was not statistically reliable, is replicated, poor readers might not be automatically sensitive to orthographic regularity.
METHOD Subjects The subjects were 22 third grade children; 11 poor readers ( 6 males, 5 females, mean age = 8.8 yr.) and 11 good readers ( 5 males, 6 females, mean age = 8.9 yr.). The Iowa Test of Basic Abilities was administered by school officials during the sixth month of the school year. The mean grade equivalent reading score from the Iowa test was 5.1 for the good readers and 2.9 for the poor readers. Apparaiz~sand Stimali The stimuli were prepared as slides and presented by a Kodak Carousel projector equipped with a tachistoscopic shutter. The slides were rear-projected onto a screen at the back of a black viewing box. Verbal response latencies were measured by a voice-operated relay. Each stimulus remained on until a response was emitted. The pictures were nine line drawings of items from three semantic categories: animals (dog, horse, pig), furniture (bed, table, chair), and modes of transportation (car, boat, plane). All letters printed inside the pictures were lower case. The interfering words were from the same three categories as the pictures but referred to different items. The words were: bird, cat, and fish; desk, lamp, and stove; train, ship, and bus. The nonword strings of letters were of two types, pseudowords and consonant strings. The pseudowords were constructed from the words by rearranging, across words, the initial letter or consonant blend of each word. The pseudowords were: fird, lat, tish, stesk, bamp, crove, shain, bip, and dus. The consonant strings were constructed by substituting consonants for vowels in the pseudowords. Procedure There were four conditions of interference in the experiment, reflecting
69
PICTURE-NAMING INTERFERENCE
the four types of strings of letters printed inside the pictures. Thus, subjects named pictures containing: ( 1) Consonants; ( 2 ) Pseudowords; ( 3) Extracategory Words, e.g., a picture of a dog containing the word "lamp;" ( 4 ) Intracategory Words, e.g., a picture of a dog containing the word "fish." Each of the nine pictures was presented twice per condition for a total of 72 trials per subject. Trials were presented in two blocks of 36 trials each. The first block contained all trials for the two nonword-interference conditions, while the second block contained all trials for the two word-interference conditions. Trials within each block were presented in one of two predetermined random orders. Each ordering was used for one-half of the subjects in each group. Subjects were tested individually. They were seated in front of the equipment and told that they were going to be tesred to see how quickly they could name pictures. They were first shown a sheet containing the nine stimulus pictures and were asked to name them; then they received nine practice trials with rhe pictures presented individually by slides. There were no interfering srrings of letters printed inside the pictures during practice. The subjects were then told that on all future trials the pictures would have letrers printed inside them but they were to try to ignore these. The block of trials for the rwo conditions with nonword-interference was presented first, followed after a 5-min. rest by the block of trials for the two conditions with word-interference. Subjects were instructed to name each picture as quickly as possible while trying to ignore whatever was printed inside the picture.
RESULTS The median response latency for each subject in each condition was first calculated. The means of the median response latencies in each condition are presented in Table 1. TABLE 1 MEANRESPONSELATENCY ( I N MSEC.)FOR GOODAND POORREADERSIN EACHCONDITION
Subjects
Consonant Strings
Pseudowords
Exrracaregory Words
Intracategory Words
Good readers Poor readers
907 988
947
1007 1085
1067 1201
980
Separate analyses were conducted on the median response latencies in the two word- and the two nonword-interference conditions. For the two nonwordinterference condirions, there was a marginally significant interaction of reading ability X conditions (Fl,20 = 2.99, fi < . l o ) . Tests for simple effects indicated that only the good readers responded reliably more slowly to pictures
70
R. E. GU'ITENTAG
containing pseudowords than to pictures containing consonant strings ( p < .05 ) . The analysis of the two word-interference conditions showed a reliable main effect of conditions (F1,?" = 50.91, p < .01) and a reliable interaction of reading ability X conditions (F1,20= 5.32, p < .05). Table 1 shows that the poor readers experienced a larger rather than a smaller interference effect than did the good readers. Tests for simple effects indicated that the difference between the incra-category word condition and the extra-category word condition was significant for both good and poor readers ( p < .01).
Drscuss~o~ The principal finding was that good and poor readers each experienced reliably more interference on picmre naming from intra-category words than from extra-category words, even though the word and picrure sets used as stimuli did not overlap. This finding indicates, most generally, that even when skilled readers have not been sensitized by the experimental procedure to the particular words used in the experiment, they still extract the meaning of the words automatically. Moreover, since even the children who were poor readers experienced this effecc, the finding confirms that poor readers are not deficient at processing familiar words automatically. The study by Guttenrag and Haith ( 1978) provided tentative evidence for an effecc of reading ability on automatic processing of nonword strings of letters. Present results replicated the earlier finding. Good readers, but not poor ones, experienced reliably more interference from pseudowords than consonant strings. Apparently, uke of orthographic regularity is so overlearned by good readers that they are sensitive to it even in a task in which processing strings of letters must occur automatically. Poor readers, on the other hand, appear to be less sensitive to orthographic regularity, a finding which may be related to their poor performance when required to decode pseudowords (Perfetti & Hogaboarn, 1975). REFERENCES DYER, F. N. The Stroop phenomenon and its use in the study of perceptual, cognitive, and response processes. M e m o ~ y& Cognition, 1973, 1 , 106-120. & ROSINSKI, R. R. Decoding. semantic processing, and reading comprehension skill. Child Development, 1976, 47, 252-258. GUTTENTAG, R. E.. & HAITH,M. M. Automatic processing as a function of age and reading ability. Child Development, 1978, 49, 707-716. NElSSER, U. Cognition and reality. San Francisco, Calif. : Freeman. 1976. PERFETI~, C. A., & HOGABOAM. T. The relationship between single word decoding and reading comprehension skill. Journal o f Educational Psychology, 1975, 67, 461469.
GOLINKOFF, R. M.,
Accepted June 7, 1979.
PICTURE-NAMING INTERFERENCE W I T H GOOD AND POOR READERS1 ROBERT E. GUTTENTAG University of Denver Strmmary.-22
third grade children of high and low reading ability (ns
= 1 1 ) were tested for their ability to name pictures while trying to ignore words or nonword strings of letters printed inside the pictures. Nonoverlapping sets of pictures and words were used as stimuli to avoid the possibility of sensitizing subjects to the particular words used in the experiment (Neisser, 1976). Both groups experienced more interference from intra-category than extracategory words, indicating that they processed the words automatically. Only the good readers experienced more interference from pseudowords than consonant strings, suggesting that poor readers are less sensitive than good readers to orthographic regularity.
In 1935 Stroop devised a task requiring subjects to name the ink colors of printed color-words while trying to ignore the words themselves. ,Skilled readers have considerable difficulty ignoring the words (Dyer, 1973), suggesting that they process isolated printed word automatically. Several recent studies employing a picture-word variant of the Stroop task have demonstrated chat even children who are poor readers experience interference from to-beignored printed words, suggesting that poor readers are not deficient at processing printed words automatically (Golinkoff & Rosinski,' 1976; Guttentag & Haith, 1978). Neisser ( 1976), however, has questioned the generalizability of findings from the Stroop task on the grounds that the use of interfering words which correspond to task-relevant responses may sensitize subjects to the specific words used in the experiment. This criticism applies as well to the studies which have used pictures and words as stimuli. The present study examined automatic word processing by good and poor readers with a procedure designed to avoid sensitizing subjects to the particular words used in the experiment. This was accomplished by using nonoverlapping sets of pictures and words as stimuli. For example, one of the distractor words used in the present study was "bird," but at no time were subjects required to name a picture of a bird. Hence, there was no reason for subjects to be sensitized to process the printed word "bird." T o determine whether subjects extracted meaning automatically from the printed words, the categorical relation between each distractor word and the picture it was princed 'This research was supported by a doctoral fellowship from the Canada Council to the author and by a grant from NICHHD ( H D 0 7 2 3 2 ) to Marshall M. Haith. The research would not have been possible without the cooperation of Charles Stanley, Principal, and the staff and students of Cherry Hills Elementarv School. Reprint requests should ~, o'f Winnipeg, be sent to Robert E. Guttentag, Department of ~ s ; c h o l o ~LJni;ersity Winnipeg, Manitoba R3B 2E9.
within was systematically varied. If subjects experience more interference from intra-category words, e.g., a picture of a dog containing the word "fish," than from extra-category words, e.g., a picture of a dog containing the word "lamp," then we may conclude that they have extracted meaning from the printed words. And since their attention is directed at picture naming rather than word processing, we may condude that they have processed the words automatically. The present study also examined the amount of picture-naming interference produced by nonword strings of letters. An earlier study by Guttentag and Haith (1978) suggested that only good readers experience more interference from pseudowords than from consonant strings. If this finding, which was not statistically reliable, is replicated, poor readers might not be automatically sensitive to orthographic regularity.
METHOD Subjects The subjects were 22 third grade children; 11 poor readers ( 6 males, 5 females, mean age = 8.8 yr.) and 11 good readers ( 5 males, 6 females, mean age = 8.9 yr.). The Iowa Test of Basic Abilities was administered by school officials during the sixth month of the school year. The mean grade equivalent reading score from the Iowa test was 5.1 for the good readers and 2.9 for the poor readers. Apparaiz~sand Stimali The stimuli were prepared as slides and presented by a Kodak Carousel projector equipped with a tachistoscopic shutter. The slides were rear-projected onto a screen at the back of a black viewing box. Verbal response latencies were measured by a voice-operated relay. Each stimulus remained on until a response was emitted. The pictures were nine line drawings of items from three semantic categories: animals (dog, horse, pig), furniture (bed, table, chair), and modes of transportation (car, boat, plane). All letters printed inside the pictures were lower case. The interfering words were from the same three categories as the pictures but referred to different items. The words were: bird, cat, and fish; desk, lamp, and stove; train, ship, and bus. The nonword strings of letters were of two types, pseudowords and consonant strings. The pseudowords were constructed from the words by rearranging, across words, the initial letter or consonant blend of each word. The pseudowords were: fird, lat, tish, stesk, bamp, crove, shain, bip, and dus. The consonant strings were constructed by substituting consonants for vowels in the pseudowords. Procedure There were four conditions of interference in the experiment, reflecting
69
PICTURE-NAMING INTERFERENCE
the four types of strings of letters printed inside the pictures. Thus, subjects named pictures containing: ( 1) Consonants; ( 2 ) Pseudowords; ( 3) Extracategory Words, e.g., a picture of a dog containing the word "lamp;" ( 4 ) Intracategory Words, e.g., a picture of a dog containing the word "fish." Each of the nine pictures was presented twice per condition for a total of 72 trials per subject. Trials were presented in two blocks of 36 trials each. The first block contained all trials for the two nonword-interference conditions, while the second block contained all trials for the two word-interference conditions. Trials within each block were presented in one of two predetermined random orders. Each ordering was used for one-half of the subjects in each group. Subjects were tested individually. They were seated in front of the equipment and told that they were going to be tesred to see how quickly they could name pictures. They were first shown a sheet containing the nine stimulus pictures and were asked to name them; then they received nine practice trials with rhe pictures presented individually by slides. There were no interfering srrings of letters printed inside the pictures during practice. The subjects were then told that on all future trials the pictures would have letrers printed inside them but they were to try to ignore these. The block of trials for the rwo conditions with nonword-interference was presented first, followed after a 5-min. rest by the block of trials for the two conditions with word-interference. Subjects were instructed to name each picture as quickly as possible while trying to ignore whatever was printed inside the picture.
RESULTS The median response latency for each subject in each condition was first calculated. The means of the median response latencies in each condition are presented in Table 1. TABLE 1 MEANRESPONSELATENCY ( I N MSEC.)FOR GOODAND POORREADERSIN EACHCONDITION
Subjects
Consonant Strings
Pseudowords
Exrracaregory Words
Intracategory Words
Good readers Poor readers
907 988
947
1007 1085
1067 1201
980
Separate analyses were conducted on the median response latencies in the two word- and the two nonword-interference conditions. For the two nonwordinterference condirions, there was a marginally significant interaction of reading ability X conditions (Fl,20 = 2.99, fi < . l o ) . Tests for simple effects indicated that only the good readers responded reliably more slowly to pictures
70
R. E. GU'ITENTAG
containing pseudowords than to pictures containing consonant strings ( p < .05 ) . The analysis of the two word-interference conditions showed a reliable main effect of conditions (F1,?" = 50.91, p < .01) and a reliable interaction of reading ability X conditions (F1,20= 5.32, p < .05). Table 1 shows that the poor readers experienced a larger rather than a smaller interference effect than did the good readers. Tests for simple effects indicated that the difference between the incra-category word condition and the extra-category word condition was significant for both good and poor readers ( p < .01).
Drscuss~o~ The principal finding was that good and poor readers each experienced reliably more interference on picmre naming from intra-category words than from extra-category words, even though the word and picrure sets used as stimuli did not overlap. This finding indicates, most generally, that even when skilled readers have not been sensitized by the experimental procedure to the particular words used in the experiment, they still extract the meaning of the words automatically. Moreover, since even the children who were poor readers experienced this effecc, the finding confirms that poor readers are not deficient at processing familiar words automatically. The study by Guttenrag and Haith ( 1978) provided tentative evidence for an effecc of reading ability on automatic processing of nonword strings of letters. Present results replicated the earlier finding. Good readers, but not poor ones, experienced reliably more interference from pseudowords than consonant strings. Apparently, uke of orthographic regularity is so overlearned by good readers that they are sensitive to it even in a task in which processing strings of letters must occur automatically. Poor readers, on the other hand, appear to be less sensitive to orthographic regularity, a finding which may be related to their poor performance when required to decode pseudowords (Perfetti & Hogaboarn, 1975). REFERENCES DYER, F. N. The Stroop phenomenon and its use in the study of perceptual, cognitive, and response processes. M e m o ~ y& Cognition, 1973, 1 , 106-120. & ROSINSKI, R. R. Decoding. semantic processing, and reading comprehension skill. Child Development, 1976, 47, 252-258. GUTTENTAG, R. E.. & HAITH,M. M. Automatic processing as a function of age and reading ability. Child Development, 1978, 49, 707-716. NElSSER, U. Cognition and reality. San Francisco, Calif. : Freeman. 1976. PERFETI~, C. A., & HOGABOAM. T. The relationship between single word decoding and reading comprehension skill. Journal o f Educational Psychology, 1975, 67, 461469.
GOLINKOFF, R. M.,
Accepted June 7, 1979.
