... 14.

13' 125

11.

•• 3'

~ .... """---------~

COMPETITlVE TOGETHER""--' COMPETITIVE .4.LONE 0-----0

NON-COMPETITIVE TOGETHER.- - - .. NOf+COMPETITlvE "'LOHE 0-- __ -<1

25~ ____ ~ ________________ ~

BLOCKS OF ONE MINUTE

Fig. 2. Mean number of responses each minute as a function of competition and coaction.

respond differently from females when alone and when together. This interaction is illustrated in Fig. 1. It can be seen that the output of females is about the same as that of males when a coactor is present. Solitary male performance is only slightly below that of coacting males, whereas solitary female performance is considerably less than that of coacting females.

Within Ss, there is an overall increase in responding as the session proceeds (F = 56.42, df= 4/288, p< .001) and a significant Time by Sex interaction (F = 16.45, df= 4/288, P < .001), since the rate of increase is less for females than for males. In addition, there is a significant interaction between time, competition, and coaction (F = 2.72, df = 4/288, p< .05). These trends are plotted in Fig. 2.

Coacting Ss who also are competitively motivated make more responses than coacting noncompetitive Ss until the last 2 min of the session. When Ss are alone, competition has little effect and may, in fact, depress performance slightly.

DlSCUSSION These results demonstrate that the males

"work hard" both alone and in the presence of a coactor, whereas the output of the females is quite low when alone. However, the presence of a coactor raises the performance of the females to the same level as that of the males. On this basis, the effort demanded by the task can be ruled out as a contributor to low female performance, thus supporting the initial proposal that females are more sensitive than males to the social situation.

There is no indication from the data that females are less competitive than males. In the coacting situation, the competitive instructions serve to increase the rate of responding of both sexes, but the same instructions have little effect when the Ss are alone. This suggests that some form of evaluation or comparison of results is

Psychon. Sei., 1970, Vol. 19 (6)

necessary to make the competitive instructions effective. It also is c1ear that if competition is required for coaction effects to occur, it need not be intense.

REFERENCES ALLPORT, F. H. Social paychology. Boston:

Houghton Mifflin, 1924. CHURCH, R. M. The effects of competition on

reaction time and palmar skin conductance. Journal of Abnormal & Social Psychology, 1962,65,32-40.

CHURCH, R. M., MILLWARD, R. B., & MILLER, P. Prediction of success in a competitive reaction time situation. Journal of Abnormal & Social Psychology, 1963, 67, 234-240.

COTTRELL, N. B. Performance in the presence of other human beings: Mere presence, audience, and affiliation effects. In E. C. Simmel, R. A. Hoppe, and G. A. Milton (Eds.),

Social facilitation and imitative behavior. Bos­ton, 1968. Pp. 91-110.

CROWNE, D. P., & MARLOWE, O. The approval motive: Studies in evaluative dependence. New York: Wiley, 1964.

DASHIELL, J. G. Experimental studies of the influence of social situations on the behavior of individual human adults. In C. Murchison (Ed.), Handbook of social paychology. Worcester: Clark University Press, 1935.

UESUGI, T. K., & VINACKE, W. E. Strategy in a feminine game. Sociometry, 1963, 26, 75-88.

ZAJONC, R. B. Social facilitation. Science, 1965, 149,269-274.

ZAJONC, R. B. Social paychology: An experimental approach. Belmont, Calif: Wadsworth, 1966.

NOTE 1. This research was supported by Grant

No. PA0332 from the National Research Council of Canada and agrant from the Department of Labour, Government of Canada.

Knowledge of intraword redundancy by beginning readers

DEBORAH LOTT, Harvard University, Cambridge, Mass. 02138, and FRANK SMlTH, Ontario institute for Studies in Education, Toronto, Ontario, Canada

Development of the ability to make use of sequential redundancy within visually presented words was measured by the difference between recognition threshold for letters in words and letters presented individually for 90 Ss (20 each [rom Grades 1 through 4 and } 0 adults). Results indicate that implicit knowledge of sequential redundancy in familiar three-letter words can be apparent even in Grade-} readers and increases slightly up to Grade 4, when approximately adult skill may be achieved.

Adult readers are able to predict successive letters of words with far better than chance accuracy (Garner, 1962, Chap. 7, 8), and can identify letters in words on less visual information than they require to identify the same letters in nonwords (Smith, 1969). This ability to make use of sequential redundancy, or sequential constraints, within words must be based on an acquired implieit

knowledge of word structure. In the study to be reported, the development of this knowledge of sequential redundancy was exarnined in beginning readers.

An earlier unpublished study by the present authors indicated that beginning readers quickly acquire a useful knowledge of sequential redundancy. In a comparison of letter identifications in words and nonwords, third graders made significantly more use of the redundancy in words than did first graders. However, fifth graders failed to show any improvement over the third graders, suggesting that a ceiling may be reached with simple material very quickly. The purpose of the present study was to examine more c10sely how children develop adult-level skilIs in the use of sequential redundancy in the recognition of familiar three-Ietter words.

The technique of both the unpublished study and the study to be reported here is to project astimulus word or letter at below contrast threshold and to gradually increase contrast intensity. The relative intensity of the stimulus when a letter is correctly identified is taken to be an indicator of the amount of stimulus information required to make that identification. The average difference for

343

any one S between the contrast level at which a letter is identitled in a nonword or in isolation and the level at which the same letter is identified when in a word is termed a mean gain. A significant mean gain indicates that the reader has acquired implicit knowledge of dependencies within words, that is, he can use information from other parts of the structure tr> identify letters in words. Further ratIonale, and data showing the effect among adult readers, are given in Srnith (1969).

Three hypotheses were tested in the present study: (1) that mean gain would increase over grade levels, (2) that ease of recognizing a letter in a word would be related to ease of recognizing other letters in that word-e.g., that the t in hat would be easier to recognize than the t in but if ha were easier to recognize than bu, and (3) th at letter confusions (incorrect responses) would be more likely to occur within vowel classes and within consonant classes for letters in words than for letters presented in isolation. These three hypotheses are predicted on the assumption that the information used by readers to identify any letter in a word extends beyond the letter boundaries to the structure available in the word as a whole.

SUBJECTS Eighty Ss were randomly selected, 20

each from Grades 1 to 4. All Ss were from the same school, except for 10 frrst graders. There was no significant difference in performance between the two groups of first-grade children, all of whom had successfully completed in kindergarten the instructional program from which the word stimuli were drawn. A "skilled-reader" comparison level of adult performance was provided by 10 staff members at the Southwest Regional Laboratory for Educational Research and Development (where the research was conducted).

STIMULI Twenty-eight test stimuli were used. The

stimuli consisted of 12 three-letter words (but, did, fun, hat, let, man, men, ran, sad, sun, wet, win) and their 16 component letters. When the letters were presented individually, the other two positions were filled by the # symbol-for example, when the component letters of fun were presented individually, they occurred as f##, #u#, ##n-in order to minirnize effects due to lateral inhibition.

Stimuli were typed in lower-case IBM Delegate type, photographed on orthochromatic rum to provide positive transparrncies with maximum contrast, and mounted for presentation by a Kodak Carousel projector.

PROCEDURE All testing was conducted in a mobile

344

laboratory at the elementary school where the children were enrolled. Adults were also tested individually in the mobile laboratory.

The stimuli-approximately 116.-in. high-were projected onto a 5 x 7 in. white card, 5 ft from the seated S in a lighted room. The light intensity (contrast level) of the display was controlled by a variable transformer calibrated in approximately 0.6-V steps. Each stimulus was initially projected as below contrast threshold, and intensity was gradually increased until recognition occurred.

The E asked S to 'report verbally when he was able to identify a letter or letters, together with the positions in which the letters occurred. E recorded the contrast level at which every letter was identified. Because of the irregular and sometimes lengthy exchanges that occurred between S and E, stimuli were projected for varying periods. However, preliminary research with this technique indicated that such random variation in exposure time at each level was not a pertinent variable. Mean gains were subsequently computed for each S by subtracting the contrast level at which a letter was identifled in a word from its identification level in isolation.

Eight trial stimuli were used to farni1iarize each S with the technique. For the first four trials, S was told whether to expect a word or a letter; for the remainder of the session, he was not told what to expect. Initially, S was asked, at each contrast level, to report what letters he saw, always noting the position (beginning, rniddle, end) of each letter reported. As S became familiar with the technique, the prompting was reduced.

RESULTS All groups tended to identify letters at

lower average contrast level in words than in isolation. The increase in mean gain with grade level occurred as predicted. Median values for Grades 1, 2, 3, and 4, respectively, were: 0.65, 0.67, 0.75, and 1.11. The difference between Grades 1 and 4 is significant by the Mann-Whitney U test, U(20,20) = 104.5, p< .01. The mean gain for adults was 0.96, which was not significantly different from that of fourth graders, U(10,20) = 73.5, p> .05. Mean gain did not vary significantly over the three letter positions. In summary, children appeared able to use sequential redundancy in the identification of letters in farniliar three·letter words even in Grade 1, and by Grade 4 achieved an adult level. These results substantiate the pilot study finding that an asymptotic performance on simple words may be reached before fifth grade.

The prediction that the identification level of a letter in a particular word would

be related to the identifiability of the other two letters in that word (rather than a constant for all the words in which it occurred) was tested with the five letters that occurred in three or more test words-a, e, u, n, t. Instances where all three letters of a word were identified at the same level were excluded, because they would have overestimated the chi-square value in the median test (Siegel, 1965, pp. 111-115) employed. This test provided a measure of whether above (or below) median performance on each of the five critical letters in any word was associated with above (or below) median performance on the two other letters in that word. Pooled data for the five critical letters and all ages yielded a chi square (1) of 88.92, p< .001. The data for each age and for each letter also yielded significant results (with p< .01 in all cases). These results support the view that information from one part of a word facilitates identification of other parts or letters of that word.

A tendency for consonants to be confused with other consonants and vowels with vowels was much more marked for the word presentations compared with single letter presentations: F(1,76) = 23.55, p< .001 for children; F(l ,85) = 32.55, p< .001 for all Ss; all other main effects and interactions were nonsignificant. The absence of a significant grade effect again supports the conc1usion that children from all age levels were using an implicit knowledge of sequential constraints in identifying letters in three·letter words. Even Grade 1 children can acquire the knowledge that three-Ietter words are likely to be in CVC form, and can make use of this knowledge in an identification task.

It is concluded that knowledge and use of sequential constraints develop early in reading, at least in familiar three-Ietter words.

REFERENCES GARNER, W. R. Uncertainty and structure as

psychological concepts. New York: Wiley, 1962.

SIEGEL, S. Nonparametrie statistics. New Y ork: McGraw-HiIl,1956.

SMITH, F. The use of featural dependencies across Ietters in the visual identification of words. Journal of Verbal Learning & Verbal Behavior, 1969,8,215-218.

WINER, B. J. Statistical principles in experimental design. New Y ork: McGraw·HiII, 1962.

NOTE 1. Tbe authors thank Bruce Cronnell and J ohn

Koehler for valuable assistance in designing and conducting the study. Tbe study reported was part of research conducted by Southwest Regional Laboratory for Educational Research and Development (11300 La Cienega Blvd., Ingiewood, Calif. 90304), supported as a regional educational Iaboratory by funds from the United States Office of Education, Department of HeaIth, Education, and Welfare.

Psyehon. Sei., 1970, Vol. 19 (6)