Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
VII
Introduction
This investigation was undertaken in an effort to arrive at a more complete description of how young children learn to speak than had heretofore Ix^en available. The acquisition of oral language, which has an amazingly rapid development during the preschool period, is perhaps the most important single aspect of the child's mental development, for he thereby acquires the basic tools of his later thought processes, as well as the abilities to acquire knowledge through understanding the speech of others and to communicate his ideas to those about him. Most of the earlier studies have been done on one or two precocious children and are subject to the bias of parents' reports. For the most part they have been mere vocabulary counts, which have been recorded with such a variety of methods that comparisons of the results have been impossible. Only a few isolated studies have been concerned with analysis of sentence formation, usually of samples of all-day conversations of superior children. Smith (5) devised a vocabulary test and made an attempt to analyze 1-hr, conversations of nursery-school children, but the groups were far from representative of the population and were not discrete with regard to age
and the amount of the sample of conversation varied tremendously from child to child. The present investigation is an attempt to add to our knowledge of the development of the language of the child, as it is found in samples of the running conversation of a large number of children selected so as to give a random sampling of the population. The function of language in the child's life, the changes in the length of sentences, the complexity of sentence structure, and the proportions of the various parts of speech that occur in the material are the chief aspects of the problem attacked.
Collection of Data
Subjects. Age.On the basis of several indications in the literature, children between the ages of eighteen and fifty-four months were included in the study. While the average child speaks a few words before eighteen months, it is not until approximately this age that word combinations begin to appear even in the most advanced children. By fifty-four months the period of most rapid linguistic development is usually over, and while the child's speech has not reached the degree of perfection found in the adult, his speech is so fluent that an accurate record of it over any extended period
* This summary of the original monograph (3) is published with permission of the University of Minnesota Press. 107
108 _
Tot
"5
re
-t
5
0
-
in
iCN'M
00
---M T O i
Ci
S3 o < 1 :
c
0
ot
z
^
oo
-r
>
&
i -1" CO C-J C^
7-i
;ic
1 3
00
* *
o "o
JO
-n j
n
5
X
o o
c a
All
3 TO
s
c
0
33
<N
O oys
fOC^M-< oo
fN t^ 1O fn (M o
(N
00 ""'
a
oys
-NM-ilN o P Q
fig
o
would tax the ability of the most rapid recorder. For the present study, therefore, 140 children were selected, 20 at each of the seven age levels of eighteen, twenty-four, thirty, thirty-six, fortytwo, forty-eight, and fifty-four months. In order that age trends might be revealed more clearly, even with a limited number of cases, these seven age groups were discrete with respect to age, each child being examined within a month and a half of the exact age at which he was classified. These differences, only four of which exceeded 40 days, usually canceled each other, so that the average age of each group differed from the exact age by which the group was known by not more than a week. Half-year rather than whole-year age groups were employed in order to give a more complete picture of the curve of linguistic development during a, period of rapid change. Sex.In the experimental group as a whole the twro sexes were about equally represented, as it included 67 boys and 73 girls. Within each age group both sexes were adequately represented, the ratio never exceeding 8 to 12 in any age group of 20 cases (see Table 1). Paternal Occupation.The occupation of the father was used as a criterion in order to secure a group of preschool children which was as nearly as possible representative of the preschool population of Minneapolis, where the study was conducted. The subjects were carefully selected so that each of the six categories in Goodenough's BarrTaussig occupational classification (2, Appendix A) would be represented in the same proportion by the children in the experimental group as they were among the adult males in the city between twenty-one and forty-five years in the 1920 census data. The matching was very close for the total group, as may be seen from Fig. 1, and
iris 1 All
uo.
109
ulation among whom language development had been investigated. Method of Observation.Each child was observed individually in his own home or in a place very familiar to him, such as a room in the nursery school or in one of the day nurseries of which the child was a member. Whenever possible, the child was alone 30 40
Pep cent 20
///////////////////?/^^
%%%%%^^
VI
FIG. 1.Distribution of cases by paternal occupation.
This is only two points higher for the group as a whole than was obtained in Kuhlmann's normative group at these age levels. Thus, the experimental group, while perhaps slightly more highly selected than a cross section of the population with regard to general intelligence, in spite of the abovedescribed precautionary measures, was more nearly representative than any previously assembled experimental pop-
in the room with the examiner, but in making home visits this condition was often difficult to control. Fifty consecutive verbal responses were recorded for each child exactly as they sounded to the experimenter. This number gave a fairly representative sample of the child's stage of linguistic development in a relatively short period of time, without tiring the child with a prolonged observation.
110
A response was considered as a separate unit if it was marked off from the preceding and succeeding remarks by pauses. A complete sentence was always considered as a single response, but a response was frequently less than a sentence. The responses were recorded exactly as they sounded to the experimenter, even in the cases of the youngest children, many of whom gave verbal responses that were entirely incomprehensible. However, these children have many sounds in their vocal repertories for which we have no adequate written symbols; hence, any attempt to record their utterances in writing is very unsatisfactory. Such responses have been treated separately, and the only accurate result they yield is the number of syllables combined per response. In some cases of peculiar articulation, the mother's interpretation was considered if it clarified the child's speech and if it was obvious that she was giving a literal reproduction and was not elaborating on what the child had said. The record did not include recitation of nursery rhymes from memory or responses uttered in direct imitation of another person. In order to overcome self-consciousness and to establish rapport, the child was shown picture books and toys which usually were quite effective as an introduction. The same set of toys was used throughout the experiment. The toys used were a little red auto, a cat that squeaked, a telephone with a bell, a little tin mouse, a music box, a small ball, and two picture books. The toys were not presented in the same order to all children. Frequently, the older children were asked whether they preferred to look at toys or picture books first. A preference was usually indicated, but if not, the experimenter presented the toy that seemed best suited to the age, sex, and apparent interest of the child. If it failed to bring forth any verbal responses after considerable inspection
Analysis of Data
In any attempt to deal with data of this sort, one is immediately confronted with a great many difficulties. Everyone who has tried to understand the chatter of little ones realizes from this casual observation how difficult it
111
of Response.The only quantitative treatment of these data that is possible is the analysis according to the length of response. The semicomprehensible responses and the incomprehensible vocalization were scored according to the number of syllables combined per response, while the comprehensible responses were scored by the number of words per response. In attempting to count the number of words in a response, many problems arise as to what is to be considered a separate word and what combinations are to be counted as one word. In order to overcome these difficulties and to standardize this part of the scoring system, it was necessary to formulate a set of arbitrary rules, as follows:
1. Contractions of the subject and predicate like "it's," "we're," "you're," etc., were scored as two words. In such cases the child speaks correctly according to adult conversational usage, which really is his only model of correct speech. Each part of the contraction is an essential part of the sentence, and if the sentence is to be considered complete, it is necessary to recognize these two parts. In order, therefore, to make the analysis according to the length of response consistent with the later analysis according to the construction of the response, it seemed justifiable to score such contractions as two separate words. 2. Contractions of the verb and the negative like "can't," "won't," etc., were scored as single words. The child who has no knowledge of how words are written does not know that "can't" is a contraction of "can" and "not." He hears "can" as one word and "can't" as another; they have different meanings and hence are probably two distinct and independent words to the child. 3. Hyphenated words and compound nouns, particularly proper nouns, which are not hyphenated but which probably function as single words and as names of single objects, were scored as one word. For example, "merry-go-round," "Mother Goose," "Betty Lou," and such expressions were scored as single words. 4. Each part of a verbal combination was scored as a separate word. For example, "have been playing" counted as three words. 5. "Lookit," which occurred frequently, was scored as one word if it was used alone and functioned simply as "look." If, however, it was followed by an object, it was counted as the two words "look at."
Functional Analysis.In consider-
ing the linguistic development of the young child, it is important to consider not only the increasing length of
112
speech units and their increasing degrees of complexity as the child grows older but also the functions of the child's speech in relation to his environment. What situations in which the child is placed bring about language responses? The young infant very early expresses himself in gestures, but when he begins to substitute verbal responses for his overt bodily responses, which overt responses are first superseded by verbal responses and in what situations does speech first appear? Piaget (4) considers this a problem of functional psychology, which indeed it is. Sometimes, he says, language conveys information, sometimes it provokes action in others, etc. This is an important aspect of the linguistic development of the young child that has heretofore been neglected. The old grammatical classification of sentences into declarative, interrogative, imperative, and exclamatory sentences is a crude analysis in this direction, which serves fairly well for written language and less well for adult conversation and is quite inadequate when applied to the speech of children. As Claparede so aptly says in the preface of Piaget's book, "In examining child thought, we have applied to it the mould and pattern of the adult mind." So often a sentence is structurally of one kind and functionally of another; there is so much overlapping of the categories; and so many subheads would have been necessary to make it meet the demands of children's conversation, that the rigid grammatical classification of sentences was abandoned in the present study. The function of the child's response in relation to his environment was considered the important thing in this type of approach. Practically the only attempt at treatment of this sort that is available in the literature is that of Piaget. This method has rilled a serious need in the present study very satisfactorily and has been
113
in the second type of adapted information, or the remarks about the immediate situation. If the succeeding remarks were, "I have a car like that," "Mine's broken," they would be additional information volunteered by the child, which was obviously brought about by the situation and associated with it, and hence belong in the third type of adapted information, or remarks associated with the situation. The fourth type of adapted information consists of irrelevant remarks; i.e., those in which the observer cannot notice any connection with previous remarks or actions. The second type of socialized speech is criticism, which includes "all remarks about the work or behavior of others, but having the same character as adapted information," according to Piaget's definition. In the present analysis, this classification was extended to include criticisms of objects as well as of persons and also complaints about situations in which the child is thwarted. In the third group of socialized speech, Piaget includes all "commands, requests, and threats." For the present purposes, this category has been extended to include all emotionally toned responses. Single-word sentences uttered with a decided emotional or commanding inflection are also included in this group. The fourth group consists of questions, by which are meant real interrogative sentences with an interrogative function and declarative sentences having an interrogative function, i.e., any remark that definitely requires an answer from the hearer. It does not include declarative sentences with a question added at the end merely for affirmation or approval of the statement and requiring no answer on the part of the hearer, as "I made it go, didn't I?" In the fifth group are included all answers that are "answers to real questions and to commands." In this
114
study the category includes all the elicited responses. However, remarks occurring in the course of conversation which have the form of answers but are answers to remarks that are not questions are not placed in this category, but rather under adapted information. A sixth category has been added to Piaget's classification to provide for the social phrases which occur only in social situations, but which the child has been taught to say parrot-fashion and which probably function as single words to the child, such as, "please," "thank you," "you are welcome," "bye-bye," etc. The last group of socialized speech is called "dramatic imitation," which consists of all talk in imitation of the conversation of adults, like imaginary imitative telephone conversations. It also includes imitation of the sounds made by animals, like "bow-wowwow," "meow," "moo-moo," and the imitation of the sounds of automobile horns, etc. Probably Piaget would put such remarks in one of the categories of egocentric speech, but, while many such remarks might be placed there, some of them are used in decidedly social situations and in social contexts, and hence this classification has been provided for them in this study.
Construction Analysis.We have
now outlined two methods of analysis of these data, both of which consider the response as a whole. The first was a quantitative analysis based on the length of response, and the second was a functional analysis of the response according to its relation to the child's environment or to the total situation. There is still another important aspect of the development of speech that must be considered. This is an analysis according to the construction of the response, which attempts to indicate the stage of grammatical complexity that the child has reached or to determine how closely his sentence structure approximates adult conversation.
115
1. Include no proper nouns. 2. Include no plural form unless the singular was not used. 3. Include all forms of pronouns. 4. Include no variants of verbs or of adjectives unless they are from a different root. 5. The same word may be listed more than once according to its grammatical use; i.e., if a word is used as a noun and also as a verb, it is included twice.
Reliability
At first the classifications described in the various analyses may appear to be somewhat subjective and hence unreliable, particularly in the functional and construction analyses. Two checks on the reliability of the method were therefore employed. In the first, scores yielded by the odd-numbered responses were correlated with the scores yielded by the even-numbered responses. These correlations, after correction with the Spearman-Brown prophecy formula, yielded a mean reliability coefficient of +.91 for length of response, indicating a satisfactory degree of consistency in the responses which the children made. The categories of the functional and construction analyses also yielded satisfactory odd-even reliability coefficients (above + .85) where data were sufficient to warrant such an analysis. It is probable that the few categories which were inadequately represented would also yield satisfactory odd-even reliability coefficients if longer samples of conversation were employed. The second approach to reliability attacked the problem of how consistently different scorers could interpret the scoring rules and classify the responses. Ten per cent of the data were therefore analyzed by four independent scorers. There was practically perfect agreement among the scorers on counting the words to determine the mean length of response, as indicated by a reliability coefficient of + .99. Again the size of the coefficients
116
appeared to be related to the frequency of occurrence of the type of response under consideration, the less frequently occurring responses yielding the lowest coefficients. In general the classifications proved surprisingly objective as indicated by mean correlations for each scorer with every other scorer as follows: Scorer A, + .83; Scorer B, + .84; Scorer C, +.79; Scorer D, +.81. The average scorer intercorrelation for all categories analyzed was +.78, but this value would have been +.88 if the work of Scorer C, who was obvi-
reliable single quantitative measure of the response as a whole is the mean number of words per response. As may be seen in Fig. 2, this measure shows a consistent increase with advance in age,
TABLE 2.MEAN PER CENT COMPREHENSIBLE RESPONSES BY CHRONOLOGICAL AGE AND SEX
24
49.0 78.0 67.0
30
93.0 86.0 89.0
36 88 0 99 3 93.0
42 95 5 99 8 97.2
48
99.3 99.8 99.6
ously less interested and less cooperative than the others, were omitted from the computations. It appears, then, that conscientious scorers can classify the responses of preschool children with a satisfactory degree of agreement.
the most rapid increase occurring between eighteen and forty-two months, with a more gradual increase after forty-two months. Girls used slightly longer sentences at six of the seven age levels. At the two-and-one-half-year level, where the only exception to this sex difference occurs, eight of the twelve girls happened to come from the lower Results socioeconomic classes, whereas five of Comprehensibility of Children's the eight boys are from the upper occuSpeech.Before proceeding with any pational levels. This age level reverses of the major analyses it was necessary the slight sex differences in favor of to separate the incomprehensible re- girls throughout the study owing to sponses from those which were com- this sampling factor, and hence the prehensible. It may be seen from the crossing of the curves here should not figures in Table 2 that children's speech be interpreted as having any comparabecomes increasingly comprehensible tive developmental significance for the with increase in age, that it is almost two sexes.* entirely comprehensible by the age of The mean length of response shows three or three and one-half years, and clear-cut differences among the chilthat this development occurs somewhat dren coming from the various socio* See chapter on "Language Development" by Dorothea McCarthy (1), for a summary table showing comparable data on length of response by C.A. and sex from ten investigations appearing between 1926 and 1941. In general the other studies tend to confirm these findings. Discrepancies can nearly always be accounted for on the basis of sampling or methodological factors.
117
differences do not stand out so clearly, however, because the groups were not discrete with respect to mental age and there was wider dispersion in mental than in chronological age. There is disagreement in the literature as to whether children who associate chiefly with adults learn to talk earlier and better than those who have
Gipb Both.
54
was made up of children of day laborers, many of whom had contacts outside the home, such as in a day nursery or institution. When the data for the three upper and the three lower occupational levels are grouped, the differences in favor of the upper socioeconomic levels are statistically significant at the higher ages. When the mean length of response was studied in relation to mental age, the same general trends appeared as in relation to chronological age. The
child associates. In order to shed light on this problem the subjects of this experiment were all placed in three arbitrary groups on the basis of the ages of their associates. Group 1 includes all children who spend most of their time with associates over twelve years of age. Group 2 includes children who, in addition to association with adults, also associate with children who are more than two years older than they. Group 3 includes all children who, in addition to association with adults,
118
also associate with children their own ages or younger; i.e., with children who are not more than two years older than they. These three groups were then considered with respect to the mean length of sentence. Each child was given a percentile rank within his own age
24-30 36-4Z 4834 Cfironoloqical aqc in montfia FIG. 3.Mean length of response by chronological age and paternal occupation. roman numerals refer to occupational categories. level, on the basis of his mean length of response. Then the median percentile rank for each of the above groups was computed. The results of this analysis were as follows: For 27 children whose associates were chiefly adults, the median percentile rank for length of response was 70; for 70 children who associated with older children, 42.5; and for 43 who associated with children their own ages and younger, 52.5. Thus, it appears that those who associate with adults only, seem to
The
from the lower socioeconomic groups where larger families affording such association are more numerous. Bilingualism.Fourteen children, or 10 per cent of the children in the experimental group, heard a foreign language in their homes. The median percentile rank for this group on the basis of mean length of response was 40, which is only slightly below the median for the group as a whole. This result is even more surprising when it is realized that nine were boys and only five were girls and
119
environment. For what purpose does the child talk? What needs does he satisfy by the use of verbal responses? In what situations are verbal responses brought forth, what kinds of responses are used in these various situations, and what changes do these responses show as the child grows
24
4Z 48
24 30. 36 42 46 54 Chronological a<yi in montfb Boys ^ Girb FIG. 4.- -Functional analysis by chronological age and sex (comprehensible responses only).
16
development as measured by the mean length of response. However, this does not mean that it is not a handicap to proper pronunciation and construction. Functional Analysis.Another aspect of the child's language which is interesting to consider, and which has heretofore received very little attention from investigators, is the function of the response in relation to the child's
older? These are some of the questions upon which this analysis is intended to throw some light. It is an attempt to give a total picture of the child's language and its development from the functional point of view, as it is evidenced in the samplings obtained in this experiment. As the usual grammatical classification of sentences into declarative,
120
interrogative, imperative, and exclamatory sentences proved inadequate and inappropriate when applied to young children's oral speech, the functional analysis adapted from Piaget's classification, outlined above, was employed. The percentage of the total number of responses falling in each category was determined for each child, and means of these percentages were computed according to age and sex groups as well as according to paternal occupation. While in the original report of this investigation this analysis was made both including the incomprehensible responses and excluding them, the present report is restricted to that based on only the comprehensible responses as shown in Fig. 4, which shows some of the relationships more clearly. Perhaps some incomprehensible responses belong in the egocentric category, but, since many others were definitely socialized as judged by their accompanying gestures and inflections, it seemed best to treat them separately. The results of the functional analysis failed to confirm Piaget's claim that over 40 per cent of the young child's speech is egocentric and that it becomes increasingly socialized as the child grows older. In fact in this study all egocentric categories together never exceeded 6.5 per cent at any age level, the average for all levels being only 3.6 per cent. This seeming discrepancy is probably much more apparent than real. It can probably be accounted for by (a) differences in definition and interpretation of terms, (6) differences in the situations in which the responses were recorded, and (c) individual differences in the characteristics of the children observed. Undoubtedly children's speech is often egocentric in that they speak about themselves and their own activities, but it is at the same time socialized in that it is directed
* Other writers who have attempted to use Piaget's classification subsequent to this investigation have not substantiated his claims. For full discussion of thisjproblem in the literature see chapter on "Language development" by Dorothea McCarthy (1).
121
age analysis, although they are less marked and the sex differences are less clear-cut. Construction Analysis.It is important not only to measure the extent of the child's vocabulary and the increase in the number of different kinds of words and their proportions in the child's vocabulary from time to time, but also to consider the child's progress
10
I Upper
When the functional analysis was conducted in relation to paternal occupation it appeared that the children of the upper socioeconomic levels use a larger percentage of adapted-information responses and ask more questions at all ages than the children of the lower socioeconomic levels (see Fig. 5). No significant trends appeared in the analysis according to age of associates. Analysis according to mental age groups showed the same tendencies as were revealed in the chronological
Locaep
in the ability to combine these newly acquired words into various degrees of grammatical complexity. Which grammatical combination appears first, in what sequence do more advanced forms appear, and at what ages do they appear? Adult conversational usage was the criterion of completeness of sentence structure in this analysis. The singleword sentence was scored as one type of complete response because of its adequacy in the situation. Omissions
122
of articles were disregarded. Omission oLany essential part of a verbal combination, such as the omission of the auxiliary, was considered an omission of the verb. In scoring the types of omission for the incomplete sentences, the simplest possible hypothetical completion of the response was used.
60 % 70 60 50
Fcincrionolicj Cbropictg
Compound .Sentence
Complex ,5gntgnce
40 30 ZO
fta ^ fa
la6oratcd Jkntenca
10,
01 60 % 50
.Simple .5gntancg3
-40
30 20 10
16
30
5b
42
.J6
42
4fl
34
FIQ. 6.Construction analysis by chronological age and sex (comprehensible responses only).
It is for this reason that no omissions of adjectives, adverbs, or other modifiers appear in the list of omissions, for they are not essential to the simplest form of a complete sentence. The semicomprehensible responses, which were few in number, were scored according to the type of omission wherever pos-
that occur with increase in chronological age much more clearly. It will be seen in Fig. 6 that the functionally complete but structurally incomplete responses constitute the major portion (75 per cent) of the comprehensible responses of the eighteen-months-old children. They decrease in relative
123
small throughout, and no sex difference is noticeable. The number of complex sentences given above is probably too small, since the sixth type of complete sentencethe elaborated sentence, which has been studied separatelyincludes the more involved of the complex sentences. The elaborated sentence group, which includes many complex sentences, first appears among the girls at twenty-four months and among the boys at thirty-six months. The group increases quite markedly with increase in age, the girls appearing slightly superior to the boys in the percentage of elaborated sentences. The remainder of the responses, i.e., those that were not included in any of the above categories, were scored as incomplete sentences and classified according to the different types of omissions. When considered in relation to the comprehensible responses only, the tendency for the incomplete sentences to decrease with advance in chronological age is apparent. Omission of the verb, i.e., some essential part of the verbal combination, proved to be the most frequent type of omission at nearly all ages, particularly in the lower age groups. Omission of the subject comes next in importance, but it is relatively less important at the lower age levels, in which the child is in the naming stage. The only other type of omission that is at all large is the omission of both the verb and the subject. None of the other types of omission occurred with sufficient frequency to indicate any trends with chronological age or sex. The girls show a clear superiority over the boys in all the items of this classification that show developmental tendencies. The children of the upper occupational groups are markedly superior to those of the lower occupational groups in all of the items of the construction analysis. The same tendencies appear in the construction analysis when it is considered in relation
124
to mental age as were found in relation to chronological age, although the sex differences are less apparent. No differences were brought out among the three age-of-associates groups on the basis of the complexity of grammatical structure. Word Analysis.In the present study, while no attempt has been made to secure a complete record of each
TABLE 3.MEAN NUMBER OF WORDS AND MEAN NUMBER OF DIFFERENT WORDS USED IN SAMPLES OF 50 CONSECUTIVE RESPONSES BY CHRONOLOGICAL AGE AND SEX
months
Boys
8.7
All
18 24 30 36 42 48 54
37.3
child's vocabulary, a sampling of each child's language has been obtained under the standardized conditions described above. A word analysis of these responses should give a picture of the composition of the active vocabulary of each age level as it is obtained in a short sample of running conversation. According to the method used in this investigation, however, the par+lcu, ar words used were determined largely by the situation. The resales <"i this analysis, therefore, do '*ot ^ive any indication of the extent of vocabulary, but rather give a pict\ re of the proportions of the different parts of speech as they enter into representative samples of t\e child's running conversation. Throughout Jie tabulation of the words, the functional use of the word was regarded as the significant aspect to consider in the classification. All responses were therefore placed in the various categories of the different parts of speech, according to how they
participles always counted as verbs except when the latter were used as adjectives. The variants of verbs, adjectives, and pronouns were not counted as separate words unless they were irregular and the roots of the words were changed. The total number of words used by each child during the observations increases markedly with age, starting with 20.3 words at eighteen months and increasing to 230.5 words at fiftyfour months, as is shown in Table 3. It will be noted from Fig. 7 that the percentage of nouns, which is 50 per cent of the total number of words used by the eighteen-months-old children, decreases in relative amount to about 19 per cent. The verbs increase from about 14 per cent to 25 per cent in the age range. Adjectives almost double in the proportions of the total number of words from eighteen to fifty-four months. Adverbs show no consistent tendency with increase in age, and they never exceed 9 per cent
125
this tendency becomes less marked at the upper ages. As may also be seen from Fig. 7, the sex differences in favor of the girls, which were noted in connection with the other types of analysis, also appear in this analysis according to the various parts of speech. It is obvious that the proportions
AdwrB
"Prgpasition
n Interjection.
Adjective
I 1
4Z.
10
Mi^callanoouJ
10
24
30
36
36
42
48
54
Fio. 7.Parts of speech by chronological age and sex (based on total number of words used in 50 responses per child).
month-old children and increase with age from about 3 per cent at two years to about 7 per cent among the older children. Conjunctions appear late, and, although they show a steady increase with age, they never equal more than 4 per cent of the total number of words obtained. Interjections decrease with age from 16 per cent of the words of the boys eighteen months old to less than 1 per cent among the children of fifty-four months. The mean number of words secured from the girls is greatly in excess of the mean number of words used by the boys at the lower age levels, but
of the parts of speech according to the number of different words is determined largely by the proportions of the different parts of speech that are represented in the dictionary. When, however, the active vocabulary is considered regardless of variety, a much better measure of linguistic development is obtained. This is seen by the fact that the changes with age appear so much more clearly in the analysis according to all the words used. It has been shown that, in considering the parts of speech as they appear in the child's vocabulary, we are deal-
126
ing with material that is rather questionable from the psychological standpoint. The question of the part of speech under which the single-word sentence is to be classed, the problem of the inclusion of variants of words, the use of the total number of words, or simply of the number of different words usedall make material of this sort rather elusive. It is well to note, however, in connection with the method used here, that essentially the same results have been obtained for this larger number of children, using only samplings of their speech recorded in a comparatively short period of time, as have been found for small numbers of children for whom very careful and detailed records were kept over long periods. It would seem from such indications as these that nearly as good a measure of the child's linguistic status can be obtained by a series of short observations as by longcontinued study.
of child psychology. Boston: Houghton, in press. Chapter onLanguage development by Dorothea McCarthy.
2. GOODENOTJGH, FLORENCE L. The
Kuhlman Binet tests: A critical study and evaluation. Inst. Child Welf. Monogr., No. 2. Minneapolis: University of Minnesota Press, 1928. Pp. via + 146.
3. MCCARTHY, DOROTHEA. Language
development of the preschool child. Inst. Child Welf. Monogr., No. 4. Minneapolis: University of Minnesota Press, 1930. Pp. xiii + 174.
4. PIAGET, JEAN. The language and
thought of the child. New York: Harcourt, 1926. Pp. xxi i + 246.
5. SMITH, MADORAH E. An investigation
Conclusion
As has been shown in the study here described, the average child of eighteen months is just beginning to use words to communicate his ideas to others. Previously, he could communicate with those around him only by crying and by gestures. By the age of four and one-half years, he is a highly social individual, using language for his every need and desire, not only physical, but intellectual, and, in addition, he is using all the most complex forms of sentences found in adult conversation and his vocabulary amounts to several thousand words. A glance at any of the charts will suffice to show that the changes that occur are very rapid up to three and one-half years of age. From then on, the developmental process continues at a slower rate for the next few years. The decrease in the percentage of the child's conversation that is incomprehensible and the increase in the mean length of response
of the development of the sentence and extent of vocabulary in young children. Univ. la. Stud. Child Welf., 1926, 3, No. 5. Pp. 92. SELECTED BIBLIOGRAPHY ANDERSON, J. E. An evaluation of various indices of language development. Child Develpm., 1937, 8, 62-68. The development of spoken language. Thirty-eighth Yearb. Nat. Soc. Stud. Educ, Chap. X, Part I, Child development and the curriculum, pp. 211-224, 1939. BATEMAN, W. G. TWO children's progress in speech. J. educ. PsychoL, 1915, 6, 475-493. Papers on language development. I: The first word. Fed. Sem., 1917, 24, 391-398.
DAVIS, E. A. The development of
linguistic skill in twins, singletons with siblings, and only children from age five to ten years. Inst. Child Welf., Monogr., No. 14. Minneapolis: University of Minnesota, 1937. Pp. ix -f 165. DAY, E. J. The development of language in twins: I, A comparison of twins and single children. Child Develpm.^ 1932, 3, 179-199.
127
LYMAN, II. L. Summary of investigations relating to grammar, language and composition. Supplementary Educational Monographs, No. 36, Chap. I l l , Chicago: University of Chicago Press, 1929. Pp. viii + 302. HEIDER, F. K., and G. M. HEIDER. A comparison of sentence structure of MARKEY, J. F. The symbolic process. deaf and hearing children (In) Studies New York: Harcourt, 1928. Pp. in the psychology of the deaf No. 1 by ix + 142. the Psychological Division, Clarence MCCARTHY, D. A comparison of chilW. Barron Research Department, dren's language in different situations Clarke School for the Deaf, Northampand its relation to personality traits. ton, Mass. Psycho. Monogr., 1940, 52, J. genet. Psychol., 1929, 36, 583-591. No. 1, 42-103. Language development. (In) HENLE, M.J and M. B. HUBBELL. Murchison, Carl (ed.). A handbook of " Egocentricity " in adult conversation. child psychology, Chap. VIII, pp. 329J. soc. Psychol, 1938, 9, 227-234. 373, 2d ed. rev. Worcester, Mass.: Clark HORN, E. The commonest words in the University Press, 1933. spoken vocabulary of children up to and including six years of age. Chap. MCGRANAHAN, D. V. The psychology of VII, pp. 186-198. Yearb. Nat. Soc. language. Psychol. Bull., 1936, 33, 178216. Stud. Edvc: Part I. 1925. HORN, M. D. The thousand and three NICE, M. M. Concerning all-day conversations. Ped. Sent., 1920, 27, 166-177. words most frequently used by kindergarten children. Childh. Educ, 1926Length of sentences as a criterion of a child's progress in speech. J. 1927, 3, 118-122. educ. Psychol., 1925, 16, 370-379. International Kindergarten Union. A An analvsis of the conversation study of the vocabulary of children of children and adults. Child Develpm., before entering the first grade. Bal1932, 3, 240-246. timore: Williams & Wilkins, 1928. Pp. 36. OLSON, W. C , and V. S. KOETZLE. Amount and rate of talking of young IRWIN, O. C. Research on speech sounds children. J. exper. Educ, 1936, 6, 175for the first six months of life. Psychol. 179. Bull., 1941, 38, 277-285. PIAGKT, J. Le langage et la penseV JERSILD, A. T., and R. RITZMAN. Aspects chez l'enfant. Neuchatel & Paris, of language development: the growth Delachaux & Niestle", 1924. Pp. 341. of loquacity and vocabulary. Child The language and thought of the child. Develpm., 1938, 9, 243-259. Trans, by M. Warden. New York, JOHNSON, E. C , and C. C. JOSEY. A note Harcourt ; London, Kegan Paul,Trench, on the development of the thought Trubner & Co., 1926. Pp. xxiii + 246. forms of children as described by Piaget. J. abnorm. soc. Psychol., 1931, RUGG, II., L. KRUEGER, and A. SONDER26, 338-339. GAARD. Studies in child personality. I. A study of the language of kindergarten LABRANT, L. L. A study of certain children. / . educ Psychol., 1929, 20, language developments of children in 1-18. grades four to twelve inclusive. Genet. Psychol. Monogr., 1933, 14, 387-491. SEASHORE, R. H., and L. I). ECKERSON. The measurement of individual differLEWIS, M. M. Infant speech: A study of ences in general English vocabularies. the beginnings of language. New York: J. educ. Psychol., 1940, 31, 14-38. Harcourt, 1936. Pp. xii + 335. The beginning of reference to SMITH, M. E. An investigation of the development of the sentence and the past and future in a child's speech. extent of vocabulary in young children. Brit. J. educ. Psychol., 1937, 7, 39-56. Univ. la. Stud. Child Welf., 1926, 3, The beginning and early funcNo. 5. Pp. 92. tions of questions in a child's speech. The influence of age, sex, and Brit. J. educ. Psychol, 1938, 8,150-171. situation on the frequency, form, and LITTLE, M. F. and H. M. WILLIAMS. An function of questions asked by preanalytical scale of language achieveschool children. Child Develpm., 1933, ment. Development of language and 4, 201-213. vocabulary in young children. IV. Univ. la. Stud. Child Welf., 1937, 13, A study of some factors influencNo. 2. 49-94. ing the development of the sentence in
128
preschool children. J. genet. Psychol., 1935, 46, 182-212. Some light on the problem of bilingualism as found from a study of the progress in mastery of English among preschool children of nonAmerican ancestry in Hawaii. Genet. Psychol. Monogr., 1939, 21, 121-284.
VAN ALSTYNE, D. The environment of
three-year-old children: factors related to intelligence and vocabulary tests. Teach. Coll. Contr. Educ, 1929, No. 366. Pp. 109.
WELLMAN, B. L., I. M. CASE, I. G. M E N GEBT, and D. E. BRADBURY. Speech
sounds of young children. Univ. la. Stud. Child Welj., 1931, 1, No. 2, Pp. 82. WILLIAMS, H. M. An analytical study of language achievement in preschool children. Part I of Development of language and vocabulary m young
sion of the Smith vocabulary test for preschool children. Part III of Development of language and vocabulary m young children. Univ. la. Stud. Child Welf., 1937, 13, No. 2, 35-46. New Series 336. YOUNG, F. M. An analysis of certain variables in a developmental study of language. Genet. Psychol. Monogr., 1941, 23, 3-141. ZYVE, C. I. Conversation among children. Teach. Coll. Rec, 1927, 29, 46-61.