The Symbolic Process and its Integration in Children

Chapter 7: The Use of Personal Symbols

John Fordyce Markey

Table of Contents | Next | Previous

To continue the study of the role of persons in the symbolic development of the child, materials on the ordinary speech activity of children were obtained. While the acquirement of a symbol may be significant, the further use of it in the behaviour processes of the child is also an important aspect of the integration of symbols.

The great amount of energy expended in speech reactions is illustrated by Table XI on all-day conversations.

The averages for ten cases are as follows :—

Years of age 3.4
Total years spoken 11,518
Words per minute 15.6

Averages for six of these cases are : —

Different words used 795.7
Use of each word 13.8

Averages for five of these cases are : —

Total percent of vocabulary 41.3
Number of sentences 2,049
Words per sentence 5.6

Brandenburg states that his 4a-year-old child was linguistically inactive only 19 minutes during the whole day of 12 hours (outside of 20 minutes in the library). These conversations appear to be taken in a home environment very favourable to speech activity. For instance, in a study of 18 children in the Merrill-Palmer school,

( 83)

Table XI
Comparison of All-Day Conversations.
Authority Child Age, Years Total


Words per minute Different Words % of Vocabu-lary Average Use of each word No. of sentences Average Words in sentence
Gale Boy C 2 10,507 15.2 805 --- 13.0 --- ---
Gale Boy S 2.5 9,290 13.4 751 52.5 12.4 --- ---
Gale Girl H 2.5 8,992 13 629 41.6 14.3 --- ---
Nice Girl D 3 7,600 10.6 --- --- --- 2,018 3.77
Brandenburg Girl G 3 11,628 16 859 34 13.7 1,873 6.6
Bell Girl A 3.5 15,230 21 --- --- --- --- ---
Nice Girl R 4 11,511 13.4 731 54.5 14.4 2,686 3.9
Brandenburg Girl G 4 14,930 20 999 24 14.9 1,967 7.5
Bell Girl B 4.5 14,992 20 --- --- --- ---
Nice Girl E 5 10,500 13.4 --- --- --- 1,702 6.17
Average   3.4 11.518 15.6 795.7 41.3 13.8 2,049 5.59
Source Nice, 1920, p. 168


(84) made by Marion Mattson (1926), the average of words per minute is lower. There were two groups of nine children in each ; the ages of Group I were from 35-40 months, average 37.4, and the ages of Group II were from 51-56, average 55.4. Excluding one case, age 37 months, in Group I, which hardly talked at all while at school, saying only 46 words in 54o hours' observation, the average for the 17 children is 6.7 words per minute. Just how much difference there is between verbal activity at school and at home it is difficult to say, but evidently there is considerable difference. If the average for the whole day were available, it would undoubtedly be much greater and would probably be still greater for children not in a nursery school. According to Miss Mattson, nursery schoolchildren are apt to talk less. The period of observation was limited in any one day to three hours.

The percentages of the different parts of speech of the conversations of some children are given in Table XII ; the prepositions, conjunctions and interjections are small in number and are not included. The figures are taken from Boyd (1914), Nice (1920), and two samples I took and classified from the all-day conversation of Brandenburg's 4o-month-old child (1915). Boyd's data were obtained by recording the sentences during the last week of the month ; the others represent all-day conversations, except that hour conversations only are reported for two of Nice's children.

Any such classification of a child's speech into the conventional forms of grammar is bound to be forced. An attempt to do it will convince the most sceptical of the truth of this statement. This fact must be kept in mind when considering the following figures. They can only be taken as indicating certain general conditions rather than precise measurements. The summary of the usage of different parts of speech (Tables XII and XIII) shows that verbs hold first place, pronouns second, and nouns third. The adverbs and adjectives are about even, with the adverbs maintaining a slightly higher rank, as will appear from an examination of the original percent

( 85)

Table XII
The Parts of Speech of Children’s Conversation in Percentages of Total Words Used
Parts of Speech Child and Age in Years
Boyd Boyd Nice Bran-denburg Boyd Nice Child R Nice R[1] Nice
2 3 3 3.5 4 4 4 5
% of Total % of Total % of Total % of Total % of Total % of Total % of Total % of Total
(1) (2) (3) (4) (5) (6) (7) (8) (9)
Verbs 27.7 22.6 23.1 26-29 21.1 28.2 29.6 28.1
Pronouns 6.2 15.8 22.3 21-22 18.1 22.1 23.1 22.8
Nouns 36.8 16.4 17.3 17 14.8 19.5 19.4 15.8
Adverbs 13.2 9.4 18.6 16 11.6 16.0 14.0 10.2
Adjectives 13.7 17.4 9.3 9 14.6 8.6 7.6 13.7
Source: Boyd, 1914; Nice, 1920; Brandenburg, 1915
1 Same child as column (7), but based upon one hour out of the day

 -ages. These data are also corroborated by Smith's study (1926) on 101 children from 2 to 5 years. Her figures are:

  Number of Children and Age Group
  19-2 yrs 28-3 yrs 32-4 yrs 22-5 yrs
Verbs 26 + 7 27 +6 26 +5 27 +6
Pronouns 16 + 6 25 +6 24 +5 25 +6
Nouns 22 +6 16 +5 15 +4 15 +5
Adverbs 21 +6 15 +5 13 +4 11 +4
Adjectives 5 +3 7 +3 11 +4 12 +5


Ranking these figures for the different ages, we obtain the following :--

  2 years 3 years 4 years 5 years
Verbs 1 1 1 1
Pronouns 4 2 2 2
Nouns 2 3 3 3
Adverbs 3 4 4 5
Adjectives 5 5 5 4

A remarkable fact brought out here is that the pronouns, although composing only a very small percentage of the vocabulary, usually around 2 or 3 per cent., hold second place in actual usage after the two-year ages.

The nouns, which usually predominate in the known vocabulary, hold third place in actual usage. The importance of action in a child's speech behaviour is again shown by the predominance in the use of verbs. The high percentage of verbs may be partially due to the structure of language. This, however, would not detract from the action character of symbolic behaviour. Language structure itself must be looked upon as being determined by symbolic behaviour, and hence its structure would reflect the action character of such behaviour. Thus, from this standpoint, these percentages would be significant as indicating the action content. It would be valuable to obtain a comparison of children's conversations with those of adults in order to discover the relative differences. "Tracy (agog, Ch. V) concludes that there is a decrease in the use of verbs in adult verbal activity and random observation would seem to substantiate this. However, more exact and reliable comparison is necessary before definite conclusions can be drawn in this respect. The higher percentage of adverbial usage at earlier ages also emphasizes the action character. If the previous inductions regarding the action content of nouns are generalized to include these data under consideration, further weight is given to the significance of action content in the child's speech behaviour.

Observations show this action content to be associated

( 87) to a very large degree with persons, indicated by the personal pronominal symbols and by other personal symbols The verbs and pronouns alone make up approximately 50% or more of the total words used by the children under consideration, with the exception of the two-year ages. The lower figure for two years is accounted for by the fact that these children are just beginning to acquire pronouns. Boyd's child is apparently somewhat atypical in this respect. While all pronouns are not personal pronouns, they are predominantly so in a child's vocabulary. Some idea of the proportion of personal pronouns to the pronouns may be obtained from Table XIV. Even at two years for Boyd's child, 61% of the pronouns were personal pronouns; at three and four

Table XIV Relation of Personal Pronouns to Pronouns
Child Age in years Per cent. Of Total Words % of personal pronouns [3] of pronouns
Pronouns Personal pronouns
Boyd 2 6.2 3.78 60.96
Boyd 3 15.8 13.34 84.4
Boyd 4 18.1 14.91 82.4
Brandenburg [1] 3 21–22 15.12 69-72
Nice D 3 22.3 “I”alone 7.67 34.4
Nice D 3   I omitted 18 times if included ... 9.67 43.4
Nice R 4 22.1 1st pp 11.76

2nd pp 2.00



Nice R 4 [2] 23.1 “I” alone 9.98 43.2
Nice E 5 22.8 “I” alone 6.19 27.1
Mattson (1926) Group 1 – 9 children, 36–20 months 15.6  
  Group 2 –9 children, 51--56 months 18.1  
1 Summarized from his data (1915).
2 Based upon only one hour's conversation.
3 In cases indicated these figures do not include all personal pronouns.

(88) years, the figures were 84 and 82 respectively. Brandenburg's 31-year-old child had about 70% of the pronouns personal pronouns.

Brandenburg (1915) published the all-day conversation of his child at 33 years. In his summary he gives only a partial list of the pronouns. In order to get the complete list I tabulated them. My figures are considerably different from his. This leads me to question his figures for both 3 1/3 and 4 1/3 years regarding pronouns. However, assuming that he made the same error on both sets of data, they are probably comparable for the comparison which is made in another connection (Table XVII, Graph 10). The figures which I obtained are used in Tables XIV and XV. Allowing for chance errors, these figures should be quite accurate.

Three cases in Table XIV show " I " alone forming from 27 % to 43% of the total pronouns. In one case, only the first and second personal pronouns compose 62%.

The percentage which the personal pronouns make of the total number of words used for the ages three years and above, Table XIV, is 13%to 18%. As Boyd's child is atypical, the 13% is undoubtedly lower than usual. Nice says that her child R used " I " for the subject in 36% of the sentences. Brandenburg says that his child at 43 used the first personal pronoun in some form in 50% of the sentences.

Further, in regard to the relation of verbs to the personal pronouns, Drevers found that of the verbs in the known vocabulary of his three children a great majority referred to the action of self or to self-the figures show 72% to 89% (Table XVI).

In checking over the all-day conversation published by Brandenburg, the information was not sufficient for one to be able to make anything but an approximate estimate; it was found that at least 8o% or more of the sentences were concerned with the self and persons, the great majority of them indicating action.

The data obtained indicate a very definite association of symbolic behaviour with personal references.

( 89)

So far we have not considered in particular another set of facts, the persons designated by nouns. The use of names of persons, such as " papa," " mamma," etc., also comprise a significant part of speech activity. In analysing Brandenburg's data, I found words for persons besides the pronouns amounted to 503. This raises the percentage of words for persons to over 19. Mattson's data show approximately 23% of the total words are personal symbols, counting double designations such as Charlie

Table XV
Brandenburg’s Child-Pronouns
Personal Pronouns Number
1st PP Sing. 1116
2nd PP Sing. 433
1st PP Plural 106
3rd PP Sing [1] 103
3rd PP Plural [1] 54
Source: Data summarized by Markey (Brandenburg 1915)
1 These contain personal pronouns designating a dog, chicken, etc. he-12, she---2, him -- 4, they -- 21, them -- 20 = 59 Forty-one of the third plural are not personal in the sense that they designated persons. A negligible number of the "you" pronouns were used to indicate other than a person. However, sociologically we know that animals and even toys are personal objects for the child.

Chaplin as only one personal symbol. Mrs. Horn's data for kindergarten children all over the country, the conversations totalling 489,555 words used, show 16.6% are personal designations, not including practically all proper names, and all words which were spoken less than 41 times. The proper names, with the exception of a few characters of more than local note, were excluded altogether, and the published list did not give words below 41 frequencies. If proper names plus the personal symbols below this frequency were included, it would undoubtedly raise the

(90) percentage to over 20. Judging from these figures, the percentage for the use of personal symbols (including pronouns) for these early ages is apparently slightly more than 20% of the total words used.

Returning to the use of the personal pronouns, their relation among themselves is noteworthy. Beginning at about 24 months, the use of them increases rapidly until they make up a good percentage of the spoken words, according to Mattson's data about 15% at 3 -}- years and about 18% at 4 + years. All but two of her cases had an I. Q. of 100 or better, most of them above this ; so that the groups are weighted somewhat in the upper half of the I. Q. scale.

In computing and comparing percentages, using the total words as a base, showing the use of the 1st 2nd, and 3rd personal pronominal symbols for Boyd's child, some significant facts came to light (Table XVII and Graph 10, Fig. A). The first personal singular pronouns held first place and increased rapidly in use until the third year, when they began to decrease relatively. The second personal pronouns started much lower, increased slowly, and were still rising at four years. The first person plural pronouns started still lower and were also increasing at four years. The third personal pronouns were somewhat below the second personal pronouns in use. The figures of Nice, Brandenburg, and also those of Drevers (Table XVI) on the verbs in relation to self also show similar tendencies, a decrease in the first person singular pronouns and a further increase in the other pronouns after three years (Table XVII, Graph 10). Nice's child R began talking very late, which undoubtedly accounts for the high figure for it at four years. On this account it might be taken as substantiating the hypothesis, for having started late (see Graph 3) it would be reaching the peak of " self " reference approximately a year later than usual. There is a substantial decrease between D, three years, and E, five years, the other two children.

It is questionable, however, whether three children, even in the same family, can be compared in this manner

( 91) with assurance that the resulting curves are typical. Consequently, both Nice's and Drevers' figures comparing these children at different ages must not be taken as having much significance unless it is assumed that each child is normal and typical in the points compared. This is probably so regarding Drevers' children; he uses them for purposes of making comparisons. From Mrs. Nice's account, her children D and E are apparently sufficiently normal in speech development for this comparison between them. Mattson's data, which I have summarized in Table XVIII and Graph 11 further corroborate these findings for a larger number of children. The " self " pronouns decrease after three years and the " other " pronouns increase.

Table XVIII The Relative Rise and Fall in the Use of the “Self” and the Continued Rise in the “Other” Pronouns
Child Age, Years Personal Pronouns –% of Total Words Used
1st Sing 2nd 1st Plural 3rd
Boyd 2 2.4 .10 .012 .00
Boyd 3 6.86 1.91 .26 1.96
Boyd 4 6.05 2.80 .78 2.21
Brandenburg 3.5 7.34 3.23  
Brandenburg 4.5 6.32 3.32 .54  
    “I” “You”    
Nice D 3 7.67 1.67    
Nice R 4 9.97 1.77    
Nice E 5 6.19 2.21    
Source: Nice, 1920; Boyd, 1914; Brandenburg, 1915.

There is considerable fluctuation in both Group I and Group II (see Table XVIII and Graph ii), so that the trend, although definite for these cases, might not be reliable if generalized to other cases. Two facts in regard to these data must be considered with reference to the validity of using them as a basis for judgment regarding other groups. (1) The number of cases is small, being


Figures A, B, C, and D

(93) limited to 18. Allowance must be made for this fact and the figures should be checked by various methods when possible. Consequently, in addition to the productmoment coefficients of correlation, the rank order coeffi-

Table XVIII Showing the Decrease in “Self” and the Continued Increase in “Other” Pronouns after 3 years
Child M.A. [3] Months Ave. Words per minute Total Different Words 540 hrs Personal Pronouns
1st Sing 2nd Per 1st Plur. 3rd Sing 3rd Plur.
Group I                
A 41 .09 25 15.2 --- --- --- ---
E 33 .59 99 7.5 1.3 --- --- ---
D 39 2.01 166 15.1 2.3 .1 .9 1.2
C 41 2.08 (162) [1] 13.6 1.6 .8 1.5 .4
B 37 3.41 231 15.4 1.1 .2 .3 .1
F 37 4.39 268 10.2 2.7 .5 .6 .0
H 50 4.78 330 12.1 2.2 .6 .0 .7
G 43 5.23 447 12.3 3.9 .7 .6 .5
I 56 6.30 440 9.6 2.9 .8 .4 .9
Average 41.9 3.21 --- 12.3 2.0 .4 .5 .4
Group II                
D 66 4.52 324 15.3 3.7 .0 .0 .0
E 70 5.89 479 10.4 3.9 .9 .9 .3
F 71 7.80 609 10.0 4.1 1.4 1.2 .6
A 60 9.07 (290) [2] 14.1 2.8 .7 1.0 .1
G 53 10.15 539 11.6 4.6 1.4 .6 .6
H 32 10.19 689 9.3 4.1 1.2 2.9 .4
B 38 11.85 (442) [2] 9.1 5.9 .8 1.0 .6
I 74 12.00 739 9.6 4.5 2.0 1.2 .7
C 69 13.34 735 11.1 4.0 1.9 1.4 .6
Average 65.9 9.42 --- 11.2 4.2 1.1 1.1 .4
Source: computed from Mattson’s data
1 360 hours
2 180 hours
3 M.A, = Mental Age

- cients p have also been computed. (2) The groups do not represent a random sample of the population. However, barring a few cases which had to be left out on account of absence or illness, these two groups include all the children in the school falling within the age limits selected. Thus,


Graph II Showing the decrease in the use of 'Self' pronouns and the increase in the use of 'other' pronouns after three years

(95) although these cases cannot be considered as random samples of the population, most of there also being average or above in mental age, still, if they can be considered as random samples or representative of similar groups at these ages, it would be legitimate to apply the laws of probability to them as representative of such other groups. It will at least be instructive to do so.

In order to test whether there is a significant difference between the mean of Group I and Group II, and thus a real trend, the Standard Error (e) was computed for the difference, for this first singular series. One case, E in Group I, was decidedly abnormal in its speech development in this respect. Consequently, it was excluded in this computation. The report from its parents said that its speech development was slow, that it did not begin to use " meaningful " words until about 24 months. It was an only child, and generally played alone while at home. This would tend to retard speech development. Its mental age was 33, with an I. Q. Of 94. In order to test the hypothesis regarding the decrease in the use of " self " pronouns after three years, it would be necessary to exclude cases of such slow speech development as well as the ages much below 36 months, on account of the fact that they might still be on the upward trend of the cycle. This case was excluded in all the computations involving the " self " symbol, but not otherwise. With the other pronouns, it would take its regular rank without danger of throwing off the computation.

The e with this case excluded is 1.15 and M1- M2= 1.7. The difference between the means is larger than the e, but if the difference is not greater than 3 x e, then this difference may be due to sampling. Consequently, we cannot be statistically sure whether this difference between the means and the consequent trend is due to chance or actually represents a general situation among similar groups. On other statistical and extra-statistical grounds, however, our conclusion is that the difference is significant. The other data, particularly those on individual cases, show the same trend. Individual variation in reaching

( 96) the peak, or the crest, of the cycle in the use of self pronouns would of itself cause a substantial amount of the Standard Error. One child would reach the high point earlier and another later than usual, thus causing variation in the downward trend regardless of a similar cycle in each case. Furthermore, the large e may be due primarily to the number of cases which, if increased, might reduce it markedly. But it seems evident that there is a great amount of individual variation between children in the time at which they reach the high point in the use of " self " pronouns. This seems to be the real explanation of a large part of the variation. For the second person, or "you" series, the figures are M1- M2 = 2.2 + e.447. This is greater than 3 x e, and is thus statistically significant. For the first plural, or " we " series, the figures are M1- M2= .7 + e..223, which is also greater than 3 x e.

Computations were not performed for the third person series. The third singular appears somewhat similar, but less consistent than the " we " series, and the third plural is apt to be unreliable, due to their usage for so-called impersonal objects. The percentages are also quite low.

In order to obtain another measure of the trend, a Pearsonian Frequency correlation between the decrease in " self " pronominal reference and mental age was computed. The results are r - - .34 + P E. .145. The r is a little above 2 x P.E. and, although not as statistically reliable as it should be, is nevertheless some indication that the r may be due to a true association to be found among similar groups. The r itself is low, however, indicating again a great deal of fluctuation in relation to mental age. But what evidence there is does confirm the hypothesis.

In order to test it further, a Pearsonian Frequency correlation between the decrease in the " self " pronominal reference and the average number of words spoken per minute was computed. The results were r = -.636 + P.E. .097. This latter r is high enough to be quite significant as well as being far beyond the P.E. Thus, in

( 97) this case, association is seen to be quite close and statistically reliable as a basis for judging this association in other similar groups.

From these correlations we conclude that there is the decrease in the use of the " self " pronouns after three years, but that, instead of being so closely associated with mental age, it is more directly associated with the average number of words used per minute, for these cases.

The question is, what is the significance of these figures ? If the use of words can be taken as a rough measure of symbolic integration, then the " self " reference decreases as symbolic integration increases. Reasons for assuming that the average number of words is such an index should be fairly evident.

The fact that the child learns words and their significance by using them and by stimulating both himself and others to respond to them shows the use of words to be very important. It seems that this might be more true of the child than the adult, because the child is just learning symbols, and their overt use must be a large factor in this process. Further weight is added to this assumption due to the fact that there is a very close association observable between the average words used per minute and the number of different words used. The correlation coefficient is r = +.954 + P.E. 0.24. Only 15 cases were used to obtain the r, as three cases out of the 18 were not observed for the same length of time as were the rest. They were consequently not comparable, nor could they be made so by a proportion ratio. It is known that the total vocabulary, as determined by vocabulary tests, correlated very highly with mental age. The total vocabulary score of a child is, according to Terman, a very good index, especially at lower ages, of intelligence level. He finds a correlation of .91 between M.A. and total vocabulary score for children from Grade I to the first grade high school. For adults the same coefficient is somewhat lower, .81. He states, " We believe it will be possible, before long, to measure the intelligence level

( 98) almost as accurately by means of 100 crucial words as it can now be measured by any existing intelligence scale " (1918, p. 464). The correlation between the average words per minute and mental age for these data is + .725, also indicating a close relationship. All of these facts tend to show the significance of this measure as a criterion of symbolic development.

It might be said that it is to be expected that with the larger number of words used, the less proportionally will be the use of " self " pronouns. But this is not true before three years, while the use of these pronouns is reaching its peak. Nor is it true for " other " personal pronouns

even after three. Further, it is exactly this point which is being emphasized ; namely, that the relative place of " self " reference does diminish with the greater use of words with greater symbolic integration, and that symbolic behaviour becomes concerned with other persons and objects to a greater and greater degree.

Further information showing the use of words to be an important criterion of symbolic integration appears in other chapters, particularly in Chapter IX on the nature of symbolic integration.

It is possible that by taking these observations in the nursery school environment, justice may not have been done to each child, and consequently more favourable conditions would give data showing even higher correlations. However, due to the fact that all figures were obtained in a similar manner, they probably do depict the situation fairly well, if not in the most accurate manner. But if a more accurate measure of symbolic integration were available, it is probable that the correlation of " self " reference with this measure would be considerably beyond - .616.

Partial correlation throws some light upon this association. To obtain a correlation between the decrease in the use of " self " pronouns and symbolic integration, as represented by the average use of words per minute uninfluenced by mental age, a partial correlation was computed by holding M.A constant. The result was an

(99) r 12.3 = - .60, which is very close to the r already obtained, and gives still further evidence of the negative association after three years of the use of " self " pronouns with symbolic integration. In the partial correlation the coefficients given in Table XIX were used.


Table XIX
The Correlation of Pronomial Usage with Mental Age and the Average Words Used Per Minute
  Series Correlated r P.E. p r [1] from p
1 Use of Self Pronouns with M.A. 17 items (2) -.34 + .145 -.55208 -.57
2 Use of Self Pronouns with A.W. per minute, 17 items -.636 +.097 -.6979 -.71
3 M.A with A.W. per minute, 18 items +.725 +.076 +.7966 +.81
4 Use of Self Pronouns with A.W. per minute M.A, held constant using r’s of 1,2,3 above -.60   -.515  
5 Use of “You” [3] pronouns with A.W. per minute, 18 items +.855 +.043 +.8751 +.88
6 Use of “We” [3] pronouns with A.W. per minute, 18 items +.85 +.044 +.856 +.87
7   (-).858      
8   (-).76 +.069    
9   (-).173 +.430    
10   .639 +.097    
11 A.W. per minute with number of different words, 15 items +.954 +.024 +.982 +.98
Source of Data Table XVIII.
1 From Pearson’s formula
2 Case E excluded
3 All forms
4 See Kelley, 1923

When looking over the frequency distribution, the correlation appears to be slightly non-linear (see Graph 11). This characteristic is further emphasized when Cases A

(100) and G are considered. Child A, according to parental report, did not begin using meaningful words until 36 months, while child G did not begin, according to parental report, until 24 months, and its I. Q. was 95, slightly below the average. It seems obvious that these two cases, as far as speech development is concerned, belong at an earlier place in the series, and are probably more nearly comparable to the higher percentages of Group I than with those of Group II. It would seem legitimate to exclude these in the computation, particularly with reference to the " self " pronouns, as it is in this series that such cases would tend to obscure the facts regarding the decrease in their use. Case E, the only other outstanding atypical case, has already been excluded when considering the " self " pronouns mainly for the reasons that its M.A. was below 33 and its percentage in connection with slow development was so disproportionate. As both of these other cases, A and G, are above 36 months, they were included in the main computations in spite of their slow development and exaggerated percentage. Later a few computations were performed excluding them for purposes of check against the figures already obtained.

A correlation for the 17 cases in this series was computed by the formulas

The resulting coefficient for hxy was (-) .858 and for hxy was (-) .639 in contrast with a linear coefficient of -.636.hxy is practically the same. Applying the correction for too fine groupings (Kelley, 1923, pp. 240-244) to hxy reduces the coefficient to (-) .76. Applying the test for linearity (Kelley, 1923, P. 238) shows the difference not to be statistically significant. But for these cases under consideration the line of x on y does drop quite rapidly and then flatten out somewhat. However, we would have to test this out with other groups before it could be decided regarding the curvilinear character of the regression.

If the correlation is curvilinear instead of linear, it is

1 h coefficients do not carry minus signs, but the positive or negative character may be determined by inspection.

(101) difficult to know just what to attribute the curvilinear nature to. It may be due to the peculiarities of the sample. It may be that after a certain reduced percentage is obtained for the self pronouns, their use remains relatively constant. Thus they would probably cease to be an index of self integration. It might be due to the fact that the child's social situation is such that there is a plateau in development. It may also be that the number of words per minute, after a certain rate is reached, represents mere verbosity and does not indicate a corresponding symbolic integration. Thus it might follow that if we could get a truer measure of symbolic integration we should still find " self " references decreasing proportionately. However, all of these hypotheses must wait for further facts.

As a slight check on the above figures, the three atypical cases, E, A, and G, were excluded in computing e, a Pearsonian frequency correlation and a curvilinear h2 for the regression of x on y. The results are as follows : M1 - M2 = 2 .25 + e 1 .066, showing greater statistical probability of significance. The r = - .667 + P.E. .096, a slightly higher linear correlation. The h2xy regression was(-) .788, somewhat lower than before. Cases A and G, when included, evidently exaggerated the fluctuation and enlarged the h1 somewhat. The correlation for too fine groupings, however, should correct for this, and it did reduce the h1 to (-) .76. This is quite close to the above h2 which, from the appearance of the frequencies, does not require much in the way of correction for too fine groupings, as they are more uniformly bunched than before. The difference between the two h coefficients and the r still suggests that further facts should be obtained to decide whether the correlation is actually more curvilinear than linear.

The coefficients of correlation between the use of " you " (all forms) pronouns and average words per minute is positive, + .855 That between the " we " (all forms) pronouns and the same variable is practically the same, + .85.

This indicates that the line of symbolic social

(102) development lies in expanding group and social integrations. Persons grow into group and wider relations. According to these data, symbolic development is at first centred around the self. This is naturally to be expected. The development is based upon the behaviour of the child in a small circle of face-to-face associations. The " you " persons are also closely associated with the " self ; " the different " you's " in the child's surroundings have their centre in the child's own " self." Greater symbolic integration brings more and more into play the " we " phases of group behaviour, which includes others with the self. This expansion goes on, the child becoming a member of different " we " groups. The third personal elements are also developing at the same time. The non- " we" groups, the "they" and "them," come into play as well. Evidently, the expansion of the personality and symbolic integration may go on until the child enters the larger universes of discourse where these more personal elements tend to be subordinated to non-personal symbolic behaviour. Also, symbolic development may come to a relative halt at innumerable points in between the selfcentred stage and an impersonal universe of discourse.

Due to the small number of cases and the chance that a large deviating item might unduly influence the Pearsonian r, the Spearman rank order p was also computed. The p merely takes into consideration the rank of the items, thus eliminating the possibility of giving too much weight to a large deviation. These p coefficients were then translated into r coefficients by Pearson's correction. The results were practically the same with two exceptions. The correlation between the use of " self " pronouns and M.A. was raised from - .34 to - .571. This also meant a change in the partial correlation coefficient of association between the Self Pronouns and the Average Words per minute with M.A. held constant. It was reduced from - .60 to - .515, which is still moderately high.

There is a hint in Mattson's data in comparison with

1 This would tend to show that there is a real association and that the' former low r was due to undue fluctuation in the data which tended to obscure the association.

(103) those of Mrs. Horn (Table XX) that probably before the age of five years-Mrs. Horn's data are for kindergarten children, and the general age of such children is from four to six--there is a relative decrease in the " you "1 along with a continued increase in the " we "1 and third personal pronouns, but more comparative data are necessary in order to judge accurately the tendencies at these higher ages. We cannot be sure that Horn's and Mattson's data are wholly comparable in this respect, although they probably do indicate the true tendency.

A summary of the frequency of the use of pronouns summarized from Mrs. Horn's and Miss Mattson's data is shown in footnote 2 (see also Tables XIV , XX). In general, the use of the personal pronouns follows the same order as the acquirement of these symbols given in the last chapter. One should take into consideration the trends which appear in them at the older ages. Thus, the conclusions drawn regarding the significance of the order of the acquirement of the personal pronominal symbols would be further

1 All forms of this pronoun.

2 The ranking in the use of the

different pronouns is :--
For Mattson's data. data.




me -- 925




us, them--189





him -- 73




myself_- 29



our >--4

self, their--3


herself, themselves -- 1

 Foy Horn's data.



my -- 6,714

we-5, 700

















Includes no frequency

 below 41

When put in comparable terms of

XX) show up more clearly.

percentage the trends (see Table


emphasized by the similar order in the frequency of the use of them. In both Mattson's and Horn's data the I pronoun alone is over 1,000 above any other word in frequency. Of course the difference is greater with the lower ages.

Table XX
Comparison of the Use of the Personal Pronouns [1]
Personal Pronouns Used Mattson’s data Horn’s data
1st sing 63.3 53.4
2nd person 21.3 13.9
3rd sing 6.0 13.1
1st plural 6.5 11.9
3rd plural 2.9 7.8
    Total 100.0 100.1
1 No word with a frequency below 41 is included. The inclusion of them should affect the percentages only slightly, on account of the large numbers and the fact that from an examination most of the pronoun, having a much larger frequency than 41, appear to be already included.

 The facts in both of these chapters concerning the appearance and use of the personal, particularly the pronominal, symbols point to the importance of the development of the "self " and of "others " in symbolic integration. It points to the symbolic process as in essence a social process of personal and group interaction and interdependence.


No notes

Valid HTML 4.01 Strict Valid CSS2