II‘IIII ' INIII 371318111“ 19")“ “SI: 11.me - II ,II1‘1g'1I‘I‘3III $III1‘Y31 IIijIJI- :I‘I‘:%"“1Tfl .. m ‘:.::‘ ml. I H . [J‘ 11‘ ‘.'“ I“ '.- '10:: I»... ‘:‘ I‘" ”'4': Q "5", I: ‘1‘: I I +11? 3'3} ‘}'1"“3I“I l :.-'.'-f' :.1- «_ WI r1“; 1;; ' ‘f p‘“? $.91; "II..1I‘ “Ag? 13.61; 1* 1 Iz--.', ..‘, WW" Whig“ 11p6¢{'1@fi:::\*:'.0 W: 1711‘} ”513‘? “' . I g“ ".13: I . 7.3-; I>q-o4 - -ur -. . 7’;— .6. fleas..- “ é':;o. 92*.“ ”€— ~ 1.1.311 I1: ‘ y; 91;. [firiwi-i“ .. . c. {17 I” 1:34” '|,"1' I if; J1EC‘1 k .I:M‘;’¢‘ $1.4” I}Lp\'lf 0. M. ‘ ~ 2%,.1. ' 1 11. V “$1111,“ '{ TIC...I-1“f.:4\ ”17'- Y‘a‘fira '1‘ ‘ ' 63111-5313‘. " -. IN: - “- I ~Y‘: 'I'I- wh31_11$1.fl\l xiii; I‘I'g “[113 ~T2pn V." fiffié { jg‘d'fizfiu ’. .. I I, l 1' “‘ ‘ ‘- v‘ ‘ . “113 :“I""',II|"11‘1"F’3\ Iggy. «$11,; 1mg; 13>: III ”5.11M“. “ . ‘ ' . 1 - . . ‘1 - ' I 1 £1“- 1‘. ‘ H“ ‘P "1" A ”W"; ‘ "1“?" 3‘1‘ ‘ ”*kang ”\hu‘I *331'8‘5' "%\. ‘ ‘2 ‘-"\I .'.‘ 1" ‘. II-‘I ‘ ’2? ~ “‘25 “733.315.321 1‘ ' 31v IL.I . ~ 1112‘}- “'51:: I 2110 I\fl!‘f‘h I." lggrg'm" :55. I: . 3.17.335“... '1‘ - n.3, 1753‘“! 1‘32“ 'IIIJH“, . “I “J" 57-5148; 'ET ~I‘ I1 , «I. .w. 1 11 “I. .1: 1' 1:"... 'I‘ r}? "f '. I‘ ’9- 31,1. J’mrg‘. “Vb“- ..JII' 1:131: ‘ " 1.1L: 1...: girl-1g”?'-=1,\II:~>.:I.1§.§‘14*1.I . I- L‘Ix'r‘ , -. -. .u. '- ,. .‘. “I U “11" "1. ‘39'11'; al.15'5-V’1911’1’w1-“I” 221:332311 ‘13:.» x I .' :II L'" "t‘ A “UV“ .. GEL“; 1,) 53.23 ‘:fi~%g-'3"n¢- I "L; Fab-*1 w’r‘. 51-11.:y'M-Jag.:.«..1:3I‘I*:147- . .' : —' ' an. ' " I "f“- ' ' ‘ I .1 ‘20:; -'§r‘§j‘rt;§2°~~?§§~2¥ ng:'v=fI;?JE-..~§i’-'." 1- - - -- :.~.- 1. 2.‘ a -" . " '3." ”.1"? 'I“ fist-$151} $131717":r13'fi.... 1.: :3: .I4 . .""I"'I.\ '~- '-‘-~" ‘ " 3-. "13:1.- u‘I‘é‘ VIII-nit“ 1‘ .9! M 'II' 1%" I.‘ I'I'1‘1. ‘1‘ ‘1. - .. Ifi'fii Id. “\V-fl - I 3‘}: 1 ' I ‘1'.-. Q I51 ll.’ I I'IIII' ‘ , ,1“. ' ' I' I'i? 3"" ' .. I I. I i‘ 1 1 I“ " :1 I II 1 I "If I . 1| *I,.,I' I .' n . “I . I ,II 1-.7- ., I , III. "LJI‘III II II._II1|I1 1I II. .I II..II I II“? Q '1) I I“ 1 " ‘1‘“ " "" ' ' 19H“ "1.. ‘. 1 .>;~.3'."‘" I;-'.I 4":2' .‘v 1“ 1.1“l|"““t“ ‘ ' ' '3 W. 1“. If. ‘, I . I I‘I'T “," 1-‘-' I‘ I ‘1' III1 . ‘J. II.""1 "II.1 I IIII‘ :1I "."‘ .I . I I-jI. .I "1 I '.I-I'.' . .I :I'Ifr Q ‘1 I .1 L" 1 6.34 I ' I? I ‘. .I-‘Io J:.. I ' ‘. IP'1.', ‘lfl vim I]? '.I‘L|-I"V“"1‘IL"‘ .‘ I" Im'I. ‘3" ‘I 113”“. I) . "“111 ';.:>1'..‘LI}‘I~’-' '- "1'5 .II HIKINF L' I‘m-“1' I‘ , l. | I I O ‘I I ‘ ‘ ‘ 'Iu'lllv.l| ‘ .. II. ""1".“ ' 3.1 "In 3'" T'nk‘ I .InIfiII 1,. .II‘ C -. 2I1‘_*.‘: ‘I I'IFS'I-‘I. .19.”. 1‘11. '4 I ‘4 11‘. 'I‘ I.» . 1 I I” I . 1' . . 1 ‘IJII3III II, . J‘III“ .12.}: .9111" ”fin-1,1...“ ~r‘~-" ,::;‘- JI‘I'1’ I o thBIS I//I/////I////II///I/////Ii’ll/I/Il/i’llI/I/lI/lf/I/fl/lII/lfl/I 3 1293 10468 3465 This is to certify that the thesis entitled EXPRESSIVE AND REFERENTIAL COMMUNICATION IN CHILDREN'S EARLY LANGUAGE DEVELOPMENT THE RELATIONSHIP TO MOTHERS' COWIUNICATIVE STYLES presented by M. Diane Klein has been accepted towards fulfillment of the requirements for Ph.D. . Psychology degree in 5/20/80 Date 0-7 639 , ITIBRARYQ Michigan Sta. .1 :2 Univuu’ty!‘ OVERDUE FINES: 25¢ per day per ite- RETUlIuNG LIBRARY “MATERIAL; Place in bookre mu charge from circulation room 62. 9 6'3: NOV {gag EXPRESSIVE AND REFERENTIAL COMMUNICATION IN CHILDREN'S EARLY LANGUAGE DEVELOPMENT: THE RELATIONSHIP TO MOTHERS' COMMUNICATIVE STYLES BY M. Diane Klein A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Psychology 1980 ylla/Nf‘ ABSTRACT EXPRESSIVE AND REFERENTIAL COMMUNICATION IN CHILDREN'S EARLY LANGUAGE DEVELOPMENT: THE RELATIONSHIP TO MOTHERS' COMMUNICATIVE STYLES BY M. Diane Klein This study investigated the relationship between mothers' com- municative styles and children's early language acquisition strategies, particularly with regard to early lexical development and language function. The subjects of the study included 19 mother—child pairs. The children were all only children from English-speaking, intact families, where the mother was unemployed outside the home. These children were followed longitudinally from their earliest vocabulary acquisition, through the first five word-combinations. Daily language diaries were kept by the mothers, and bi-weekly home visits were made by the investigator. When each child had acquired approximately 30 words, a voice-actuated recording system was placed in the home for two days. The data collected for this study were of two types. First, the first 50 words acquired by each child were recorded and analyzed as to form and function of each word. Second, the tape recorded material ob- tained via the two-day audio recordings was inventoried and catalogued according to key activities and interactions, e.g., mother-child play, caretaking activities, father-child interaction, meal times, etc. On the basis of the number of "general nominals" in the child's first 50 words, each child was categorized as having a predominantly M. Diane Klein "expressive" or "referential" style of language acquisition. Several differences were found between the two groups of children. Referential childrens vocabularies consisted mostly of names for things. Expres- sive children's vocabularies contained relatively more personal-social expressions, and a greater variety of word categories. Language function also differed significantly for the two groups. While referential children used language primarily to label or comment on their environment, expressive children were more likely than referential children to use language to control their environment and to get attention. Mother-child interaction styles were compared for mothers of expressive and mothers of referential children. Two mother-child interaction contexts (Breakfast and Play) were selected from the second day of taping for each of the ten children. A total of 22 measures were obtained from the verbatim transcripts of the mothers' child-directed utterances in the two situations. The design consisted of two independent factors: Group (referential and expressive) and Situation (breakfast and play) with a repeated measure of the situa- tion factor. Several differences between the two groups of mothers were found. Referential group mothers used more nouns, more object references, more imitations, more yes/no questions, and fewer commands and pro- hibitions, than did expressive group mothers. Mothers of referential children also played more frequently with their children than did M. Diane Klein mothers of expressive children. A positive correlation was found between education level of parents and degree of referentiality in children. To Marv and Erin ii ACKNOWLEDGMENTS A very special acknowledgment must be given to Professor Helen Benedict, without whose support and positive reinforcement I would never have completed this degree. In addition to her encouragement and support, her thorough and critical evaluation of my work always pro- vided the direction I needed to keep moving forward. I also wish to acknowledge the members of my committee, for their input and support: Professor William Crano for his statistical consul- tation; Professor Hiram Fitzgerald for guidance through my graduate program and for providing me with a basic understanding of child development; Professor Julia Falk whose teaching inspired my interest in child language acquisition, and whose critical feedback on my re- search was timely and invaluable. Special appreciation is extended to the parents of the children who served as subjects for this research. Their cooperation and en- thusiastic participation made this study possible. Finally, a very sincere expression of gratitude must be extended to Erin and Marv: Erin, who was my inspiration, and Marv, whose con- tinued support, and sharing of the responsibilities of parenthood, enabled me to do this research. iii TABLE OF CONTENTS List of Tables. . . . . . . . . . . . . . . . . . . . . . . . . . . vii List of Figures. . . . . . . . . . . . . . . . . . . . . . . . . . viii INTRODUCTION. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Background. . . . . . . . . . . . . . . . . . . . . . . . . . . .1 Statement of the Problem. . . . . . . . . . . . . . . . . . . . 10 Research Questions. . . . . . . . . . . . . . . . . . . . . . . 14 METHOD. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .16 Research Design. . . . . . . . . . . . . . . . . . . . . . . . .16 Subjects. . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 Procedure. . . . . . . . . . . . . . . . . . . . . . . . . . . .19 Part I: Determining Child's Communicative Style. . . . . .19 Part II: Determining Mother's Communicative Style. . . . .20 Equipment. . . . . . . . . . . . . . . . . . . . . . . . . . . .23 Measures. . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 Child's Communicative Style. . . . . . . . . . . . . . . . 25 Nelson's Categories. . . . . . . . . . . . . . . . . .25 Function Categories. . . . . . . . . . . . . . . . . .26 Mother's Communicative Style. . . . . . . . . . . . . . . .29 Reliability of Measures. . . . . . . . . . . . . . . . . . 35 RESULTS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 Children's Communicative Styles. . . . . . . . . . . . . . . . .37 iv Group Comparisons. . . . . . . . . . . . . . . . . . . . . 42 Nelson's Categories. . . . . . . . . . . . . . . . . .43 Function Categories. . . . . . . . . . . . . . . . . .47 Mother's Communicative Styles. . . . . . . . . . . . . . . . . .51 Group Comparisons. . . . . . . . . . . . . . . . . . . . . 56 Noun-Pronoun Use. . . . . . . . . . . . . . . . . . . 56 Repetitions. . . . . . . . . . . . . . . . . . . . . .63 Imitation and Yes/No Questions. . . . . . . . . . . . 63 Utterances of Five Words or More. . . . . . . . . . . 64 Commands and Prohibitions. . . . . . . . . . . . . . .64 Object References. . . . . . . . . . . . . . . . . . .65 Non—significant Measures. . . . . . . . . . . . . . . 65 Summary of Differences in Mother's Communicative Styles. . 65 Differences in Two Situations: Breakfast and Play. . . . .66 Noun-Pronoun Use. . . . . . . . . . . . . . . . . . . 66 Repetitions. . . . . . . . . . . . . . . . . . . . . .67 Commands and Prohibitions. . . . . . . . . . . . . . .67 Object References. . . . . . . . . . . . . . . . . . .68 Summary of Situation Effects. . . . . . . . . . . . . . . .68 DISCUSSION. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .70 Children's Communicative Styles. . . . . . . . . . . . . . . . .71 Mother's Communicative Styles. . . . . . . . . . . . . . . . . .73 Referentiality. . . . . . . . . . . . . . . . . . . . . . .74 Play. . . . . . . . . . . . . . . . . . . . . . . . . . . .74 Parent Education Level. . . . . . . . . . . . . . . . . . .76 Conclusions and Implications For Further Research. . . . . . . .82 Appendix A Letter to Mothers and Phone Interview Data Sheet. . . . . 84 V Appendix Appendix Appendix Appendix Appendix Diary Procedures. . . . . . . . . . . . . . . . . . . . .86 Activity Summary Sheet and In-Home Recording Procedures. 88 Transcription Procedures. . . . . . . . . . . . . . . . .90 Coding Procedures for Mother's Communicative Style Measures. 0 O O . C C C U . O O O C O D C C I C O C O O .94 The First 50-Nord Vocabulary For Each of the 19 Subjects 97 vi LIST OF TABLES Table l Coding Reliability for Measures of Mother's Communicative Style and Children's Word Categories. . . . . . . . . . . . 36 Table 2 Distribution of First 50 Words Among Word Categories For All 19 subjects. I O O I O O O O O O O O I I O O O I O .41 Table 3a Differences Between Expressive and Referential Groups in Number of Words in Each Nelson Category. . . . . .44 Table 3b Differences Between Expressive and Referential Groups in Number of Words in Each Function Category. . . . .48 Table 4 Average Number of 30-unit Recorder Intervals in Which Play or Book Reading Occurred. . . . . . . . . . . . .54 Table 5 Mean Scores for Mother's Communicative Style Measures by Group and Situation. . . . . . . . . . . . . . .60 vii Figure 1 Figure 2 Figure 3 LIST OF FIGURES Definitions of Word Categories Used to Describe Children's First 50 Words. . . . . . . . . . . . . . . . . 39 Examples of Multi-word Strings Containing Pronouns of Function Words. . . . . . . . . . . . . . . . . . . . . 46 Summary of Measures Used to Evaluate Mother's Communicative Styles. . . . . . . . . . . . . . . . . . . .57 viii INTRODUCTION Background During the 1960's, research in the area of child language was directed primarily toward the discovery of linguistic universals, or characteristics of language develOpment which were common to all children. Even those researchers who were not tied to the Chomskian notion of innateness were guided by the search for commonalities as they described the language of young children (e.g., Brown, 1973). These early descriptive studies produced different characterizations of the nature of early language development, for example, Braine's "pivot-open grammar" (Braine, 1963), and Brown's "telegraphic speech" (Brown and Fraser, 1963). There was little consensus during this early period regarding the characteristics of child language development. This lack of consensus was not surprising in light of the very small samples used. Recent studies by Bloom (Bloom, 1970; Bloom, Hood and Lightbown, 1974; Bloom, Lightbown and Hood, 1975) and Nelson (1973) have suggested that there is 223 one universal formula which characterizes all language develop- ment. Rather, there are important differences among individual children which do not simply reflect population variance, but which may result from different "language acquisition strategies" (Bloom, Lightbown and Hood, 1975) or "cognitive styles" (Nelson, 1973). During 2 the early years of language acquisition research, the use of small samples, together with the strong theoretical biases of innateness and universality, would have masked such differences in patterns of acqui- sition. Nelson (1971, 1973), who investigated 18 children longitudinally from one to two years of age, concluded that there were two styles of early vocabulary development which she labeled "expressive" and referential." The early language of referential children was charac- ' or words which terized by a greater proportion of "general nominals,’ labeled object classes (e.g. "dog," "book"), while the early vocabu— lary of expressive children was characterized by a greater variety of word types, and significantly greater use of stereotyped social expressions (e.g. "thank you") and function words (e.g. "where"). Also, expressive children were more likely to produce phrases earlier, and used a greater proportion of pronouns. Thus, while the typical referential child's vocabulary consisted primarily of names of ob- jects, the expressive child's vocabulary was more apt to include a greater proportion of such utterances as "night night", "thank you," "mine," "oh oh", etc. In terms of differences in cognitive schemata, Nelson's findings suggest that the referential child is very object- oriented, while the expressive child seems more aware of his/her own "actions and affectivity." Lois Bloom's work has also revealed different patterns of language acquisition. Bloom's 1970 study of the syntactic develop- ment of three young children did much to refute notions about uni- versal structure in beginning word combinations which had been sug- gested earlier (e.g. McNeill, 1970). Bloom studied three children 3 longitudinally. Her study suggested that, while there were certain consistent similarities among all three children, they seemed to differ in their acquisition strategies for certain aspects of linguistic development. Subsequent work by Bloom has investigated these differ- ences in more detail. Bloom et a1. (1974) considered the role of imi- tation in language acquisition and found that children could be cate- gorized as imitators or non-imitators, and that children who did imi- tate were clearly using imitation as a language acquisition strategy. In another study (Bloom et al., 1975) Bloom examined differences in patterns of structures in the emerging grammar of four young children. While she found a general pattern among the four children in the development of verb relations categories (verb relations being the primary focus of the study), she also found that children used very different linguistic systems for representing essentially the same information. According to Bloom, there appeared to be two mutually exclusive systems of semantic-syntactic structure. Two of the children had grammatical systems characterized by a predominance of nominal forms, and few pronouns or function words, while the other two children used more verb forms together with constant functional forms such as "I," "it," "there," "my," etc. The similarities between these findings and Nelson's expressive and referential communicative styles are striking. It should be noted, however, that the two authors differ in the following way: Bloom suggests that, while children use dif- ferent features of the linguistic code in the development of their systems of semantic-syntactic structures, they all talk about the same things. Nelson, on the other hand, implies that the early lin- guistic differences among children result from different ways of 4 conceptualizing the world and do reflect differences in what children choose to talk about. Other authors, who were not primarily concerned with individual differences, have, nevertheless, reported differences which seem to reflect the expressive-referential distinction. Dore (1974), in describing the early pragmatics of two young children, referred to one child as "code oriented," and the other as "message oriented." The "code oriented" child talked primarily about the things in his environment, while the "message oriented" child used language to mani- pulate peOple. Vander Geest (1977), who studied Dutch mother—child language interaction, commented that two of the children in this study showed differences very similar to those reported by Nelson. Lieven (1978) also described striking differences in the language deve10pment of two children. While one child was, in Nelson's terms, highly referential, the other child rarely referred to the immediate context, and used language more to gain attention and express needs. Lieven goes on to discuss the possible effects of these different styles upon the mother's interaction with her child. Bowerman (1976) considers the early language deve10pment of her own two children as well as the early diary information on Leopold's Hildegaard and Bloom's Allison. Bowerman points out differences in object word stability among these children and acknowledges that cer- tain children emphasize function words while others use mostly object words. She goes on to suggest that it is unclear whether this differ- ence reflects differences in the particular experiences which are conceptualized by the child or differences in what the child chooses to communicate. Bowerman (1978, p. 360) states, " . . . evidence is 5 accumulating that Nelson's Referential versus Expressive distinction may reflect the effects of genuine differences of cognitive style on children's language acquisition." The early research on child language acquisition not only ex— cluded individual differences as a topic of interest, as discussed above, but also was relatively unconcerned with the child's linguistic environment. This, of course, was due to the assumption that language acquisition was an innate process, and, that it had to be innate because the learning skills of the 18—month-old were quite inadequate for deriving complex grammatical rules from an on-going stream of adult speech. Subsequent research in the early seventies (Drach, 1968; Broen, 1972; Snow, 1972), however, was able to demonstrate that adult- to—child speech differs in very significant and predictable ways from adult-to-adult speech. Broen's study, for example, found the follow- ing: 1) Mean utterance length and speech rate were greatly reduced as compared to adult-adult speech, and increased with age of child. 2) Pauses were found to occur only at sentence boundaries. 3) Mothers made frequent use of repetition. 4) Sentence structure was simple; complex linguistic structure was not used. 5) Prosodic features were distinctive, including higher fundamental frequency of pitch, and exaggerated intonation patterns. Since this early study by Broen, the literature on mother-child language has steadily grown and many of these findings have been replicated (e.g. Holzman, 1974; Lord, 1976; Phillips, 1973; Snow, 1972; Newport, 1976; Moerk, 1974). It has also been found that non- mothers (Snow, 1972), older children (Shatz and Gelman, 1973; Sachs and Devin, 1976) and fathers (Berko-Gleason, 1975; Giattino and 6 Hogan, 1975) also modify their language in various ways while address- ing young children. Studies by Lord (1976), Moerk (1974) and Newport, Gleitman and Gleitman (1977) have stressed the highly redundant and referential nature of mother-child language. More recently, Snow ' conversation-like, vocal/verbal (1977) has reported a "turn-taking,' interaction between mother and child which begins early in infancy. Thus, it has been clearly established that there is an identi- fiable "speech register" (labeled "motherese" by Newport (1976) ) which is used when addressing young children. Subsequently, attempts have been made to examine the relationships between mothers' language and the child's language development. Researchers in this area were further motivated by results of more global studies of mother-child interaction which reported strong relationships between the quality of verbal stimulation provided for the child, and later competence, particularly linguistic competence. One such study was a study re- ported by Clarke—Stewart (1973) which investigated several mother- child interaction variables. Subjects for the study were 36 mother- child pairs; children ranged from nine to 18 months of age. The study was longitudinal, including twelve monthly home visits. One of the strongest relationships reported by this study was a positive correla- tion between amount of verbal stimulation directed to the child, and later development of cognitive and linguistic skills. Several recent investigations have tried to determine relation- ships between the specific characteristics of mothers' verbal input to the child, and the child's linguistic development (e.g. Lord, 1976; Newport et a1., 1977; Furrow, Nelson, and Benedict, 1978; Cross, 1977, 1978). One of the important underlying questions of these studies is 7 the extent to which the child's language behavior is shaped by his/her language environment. Although the methodologies of these studies (which have used simple correlational techniques) have not allowed demonstration of casual relationships, the authors have drawn various conclusions. In a study by Lord (1976), which was a two—year longitudinal in- vestigation of three mother-child pairs, a variety of correlational measures of mother and child language were made. These included such variables as mean length of utterance, sentence complexity, sentence well-formedness, sentence types, etc. The results of the study sug- gested that, in general, one could not predict specific mother measures from child measures. For example, it was not predictable that the mother of the child with the longest MLU would also have a long MLU. Furthermore, Lord found an abrupt change in mothers' language when the child began using single words, characterized by greater redundancy, increased referentialness, and shorter MLU (as did Snow (1977)). Lord therefore concludes that it is the degree of "tuning" to the child's cognitive level that facilitates language development. That is, the mother does not_£gggh language to the child, but, rather, responds to the child. The extent to which mothers' responses are finely tuned to what the child is attempting to communicate, or to what the child can comprehend will determine the degree of facilita- tion of his/her language development. Similar findings have recently been reported by Cross (1977, 1978) who found that accelerated syntac- tic development in children was related not to mothers' syntax, but rather, primarily, to mothers' use of semantic extensions. There were, however, other interesting results in Lord's study, 8 which she chose not to stress in her discussion. Surface structure characteristics of children's speech often reflected mothers' surface structures in various ways. For example, the child whose mother used the most single words and sentence fragments, also used primarily single words and fragments; the child whose mother used more complete sentences, used telegraphic sentences rather than single words or fragments. Thus, Lord's results seem to support two notions: first, that mothers are "tuned in" and very responsive to their child's level of cognitive development; and, second, that the particular linguistic style of the mother does influence, in some ways at least, the lin- guistic development of the child. A similar study was done by Newport et a1. (1977), which studied 15 mother-child pairs over a much shorter time period (two sessions, six months apart). Despite a number of methodological differences, the results of this study were strikingly similar to Lord's. NeWport et al. also conclude that a mother is not teaching her child language, since the general characteristics (e.g. sentence complexity and MLU) of the child's emerging competence seem to be insensitive to the maternal speech environment. Here again, however, as with the Lord study, such a conclusion is not fully supported by the results. Newport et al. also found that particular surface structure characteristics of the mother's language system seem to predict the rate at which the child learns certain syntactic constructions. A recent study by Furrow, Nelson, and Benedict (1978) also attempts to look at this relationship between mother language and child linguistic development. Furrow et al. also use correlational 9 techniques; but unlike the previous studies they attempted to examine the relationship between mother language and subsequent linguistic deve10pment in the child, as well as concurrent functioning. The study analyzed samples of mother and child language when the child was 18 and 27 months of age. Measures used were a subset of those used pre- viously by Newport et al. Correlational techniques analyzing the re- lationship between mothers' language when the child was 18 months old and the child's productive language at 27 months produced a number of strong correlations. Furrow et a1. concluded that the complexity of mothers' speech at 18 months was negatively correlated with later language development, and that use of certain kinds of constructions (particularly yes/no questions) related positively to later language development. Thus, contrary to the studies by Lord and Newport et a1., Furrow et a1. conclude that mother language_i§ a teaching language. Their findings would seem to support the notion that mother language shapes linguistic development in the child rather than being merely respon- sive to it. A recent article by Slobin, (1975) which reviews a number of studies investigating the characteristics of child-directed language concludes the following (p. 295): It is clear, however, that the thread thrown out to guide LAD and LAS through the labyrinth of language is sub- stantial enough to give them a start through the maze. Slobin (aside from his metaphors and his tribute to feminism!) is clearly conceding here that there is strong evidence that "motherese" can have a facilitative effect upon child language development. 10 Statement of the Problem The discussion above suggests that there is a growing interest in the effects of the linguistic environment on the language deve10pment of the child. Despite this interest in the child's natural environ- ment, the data collection procedures utilized by nearly all the studies in this area have involved tape recorded sessions, with the experimenter present, in which the mother was encouraged to "interact normally" in a play situation with her child. The earliest studies of this type were done in laboratory playrooms (Phillips 1973; Snow 1972; Broen 1972). There is some question as to just how representative such play interaction really is. White's (1971) longitudinal study of children from birth to three years reported that the typical inter- action between mother and child lasted approximately 30 seconds, and that extended interchanges of even five or ten minutes were markedly rare. Clarke-Stewart (1973) in her study of mother-child interaction, reported that an average of only 4% of a child's awake time was spent playing with the mother. In addition, the presence of an observer, even one with whom the mother has become familiar, significantly alters the context in which communication takes place. There is fairly strong evidence in the literature (Benedict, 1978; Broen, 1972; Shatz and Gelman, 1977) that context alters mothers' communication style. Furthermore, being ob- served especially alters mothers' communicative styles. Zegiob and Forehand (1975a) observed mothers and their five-year-olds under two conditions: one in which they were informed they were being observed via one-way mirrors and the other in which mothers believed themselves to be unobserved in a waiting room. Comparison of mothers' behavior 11 under the two conditions revealed that the informed condition produced more playing interactively, more positive verbal comments on the child's activity, and more attempts to structure the environment, using commands and questions. In a similar study by Graves and Glick (1978), mothers in the informed condition verbalized more, used more interrogative forms, were more positive, and used more repetition and gesturing. Labov (1972), an investigator in the field of sociolinguistics, refers to the "observer paradox" which states that the presence of an observer may so alter the linguistic behavior of the observed that it becomes unnatural and atypical. Wells (1977) reported that the mothers in his study used expansions frequently with their children when being observed because they were being observed. They were at- tempting to make the child's utterances clear to the observer. There is also evidence that children's language is influenced by context. For example, a study by Kramer, James, and Saxmon (1979) compared language samples of 5—year-olds obtained at home by the mother, with samples obtained in a speech clinic setting. Two measures were used to evaluate the children's language samples: Mean length of utterance was used to place the children in one of Brown's linguistic stages, and the Lee Developmental Sentence Scoring Test (DSS) was used to assess language complexity. When the clinic sample was used as the basis of analysis, both the MLU and D88 measures underestimated the child's ability as compared to the sample obtained in the child's home environment. It would seem that before we can meaningfully determine the re- lationships between mother and child language, we must know the 12 contexts in which mothers actually communicate with their children, as well as what they choose to talk about in those contexts, and we must be able to reliably describe the linguistic features of that communi- cation. There is little evidence that this has been done. More recent investigators 2323 attempted to increase the naturalness of the obser- vation situation by going into the child's home. However, in some cases mothers were asked to play with the child in ways that are per- haps not at all "natural," e.g. using unfamiliar, "standard" sets of toys, for extended periods of time, while the investigator records the interaction (e.g. Lord, 1976; Cross, 1977, 1978). In other studies mothers are asked to engage in their "normal routine" for one to two hours while the investigator observes (e.g. Bloom, 1970' Moerk, 1974). For several studies reported in the literature (e.g. Holzman, 1974; Drach, 1968; Lieven, 1977; Newport, 1977) detailed information on the data collection procedures is not available. Thus, while there is a growing interest in determining the £213: tionship between various aspects of mothers' communication and lan- guage development in the child (Lord, 1976; Newport, 1976; Newport, Gleitman and Gleitman, 1977; Cross, 1978), there has been much less concern about what the true nature of mother communication really is. In a certain sense, researchers in this area have put the proverbial cart before the horse. The important first question should be "what is the nature of the linguistic environment of the child?" This general question subsumes such questions as Hwhp talks to the child?" (The current emphasis on "motherese" seems to assume that other persons in the child's environment are relatively unimportant as sources of lan- guage stimulation.), and "In what contexts do these persons talk to 13 the child?" (There has been almost no investigation in this area. This is due perhaps to serious methodological problems, which will be con- sidered later in this discussion.) Only after the linguistic environment has been described can the second important question logically be asked: "What is the relation- ghtp between linguistic environment and language development in the child?" This question includes not only the relationship between specific linguistic and pragmatic characteristics of mothers' lan- guage to child language, but also the following kinds of questions? "Do the number and types of speakers in the child's environment in- fluence the way in which he develops language?" "Do the types of con- EEEEE in which language interaction takes place affect child language development?", and if so, "Does this reflect the effect of context on the linguistic characteristics of mothers' language?" Most studies investigating the relationship between mother and child language have focused primarily upon the child's EEEE of lan- guage development and levels of syntactic complexity (Cross, 1977, 1978; Lord, 1976; Newport, 1976). These studies appeared to be search- ing for what might be called "universal facilitative effects" of mother's language interaction style upon linguistic (primarily syntac- tic) development in children. As discussed above, the results of these studies have been, in some ways, inconclusive. If, as Nelson and Bloom suggest (see also Bowerman, 1976, 1978; Lieven, 1978), there are qualitative differences in children's strategies or styles of language acquisition, it might, then, prove useful to try to determine the re- lationships between particular characteristics of mothers' language and the type of language acquisition strategy used by the child. While 14 tgtg of language development is a measure which is confounded with rate of maturation, it is possible that a more "micro" analysis, taking into account stylistic differences, might lead to better under- standing of the relationship between the child and his/her linguistic environment. For example, it may be the case that children's early language learning strategies reflect some inherent cognitive style differences. If so, it is possible that as Nelson (1973) suggests, the extent to which a mother's communication style is congruent with the child's own ways of perceiving and organizing his experience will determine the degree of facilitation of language learning. On the other hand, a child's early choice of communication strategy may re- flect the particular ways in which the adults in his/her environment interact with that child. Rather than simply viewing children as either a) possessing universal language acquisition devices which will function in a fairly predictable way, regardless of environment, or b) waiting passively to be influenced by their linguistic environment, examination of individual differences may have greater potential for understanding the interactive relationship between the language learn- ing child and his/her environment. Research Questions The study to be presented here has attempted to look at some of the questions mentioned above. Specifically, this study has attempted to explore the following three questions: 1. What are the characteristics of mothers' communication to their children within the various contexts which naturally occur in the child's daily environment? Can different patterns of communication style among mothers be identified? 15 2. Do the differences between children in early language acqui- sition simply reflect random population variance, or are there identi- fiable patterns? More specifically, can a) the expressive-referential distinction as posited by Nelson (1973) be verified? That is, is it possible to identify two different strategies of early lexical development: one characterized by a predominance of labels for things in the environment (referential) and another characterized by a more even distribution of words among word categories and greater use of personal social expressions, function words, pronouns, and early use of word combinations (expressive)? And, b) can different patterns of language p§g_be identified. In particular, are there differences in the extent to which children use early vocabulary to label, to respond to events and attributes, and to make requests. 3. Given that different language acquisition patterns can be identified in young children, what is the relationship between these characteristics and the characteristics of mothers' communication styles. 16 METHOD Research Design There were two major objectives of this study. The first was to obtain an accurate record of the child's earliest use of productive language. There were a number of methodological impediments to obtain- ing this first goal. The nature of the twelve to twenty-month-old human organism is such that meaningful observations of its function outside the caretaker-child dyad are almost impossible to obtain. The uniqueness and exclusivity of the parent-child relationship--and also the very real problem of speech unintelligibility at this early age-- necessitated the use of parents as informants, thus introducing a degree of variability into the data collection process. Twenty families were selected for participation in the study. They were requested to keep careful records of their child's earliest productive vocabulary. In an attempt to reduce data collection vari- ability as much as possible, all parents received a standard set of instructions for filling out the daily diary sheets (see below). In addition, each family was interviewed by-weekly by the investigator in an attempt to insure accuracy and completeness of information ob- tained regarding the child's vocabulary and language function. The sample size originally chosen for the study was 20. While 17 this is a small number statistically, it was considered the maximum size sample on which a single investigator could carry out a longi- tudinal study, as well as subsequent data analyses. The second goal of the study was to describe the "typical" lan- guate environment of the child during this early stage of development. This required observing, in some way, each of the twenty families in their natural daily routines. As discussed earlier, few studies have seriously attempted to maximize the "naturalness" of situations from which language samples were obtained (with the exception of Wells, 1976 and Friedlander, Jacobs, Davis and Wetstone, 1972). The obvious problem here, of course, is that once any kind of observation is at- tempted, the situation is, by definition, no longer "natural." Studies by Zegiob, Arnold and Forehand (1975) and Graves and Click (1978) have clearly demonstrated observer effects upon mother-child interactions. It was, therefore, felt that, despite methodological difficulties, such effects should be minimized as much as possible. The method chosen for this purpose-—the placement of voice-actu- ated recording equipment for a two-day period in the home--admitted1y sacrifices important contextual information which could only be ob- tained visually. However, the rationale here was that contextual details which are generated by artificial and atypical interaction situations are not preferable to auditory data alone when the purpose of the study is to investigate "typical" mother—child language inter— action. Thus, the research design of the present study was multifaceted, including 1) data obtained longitudinally via diaries and interviews on the early productive language used by children; and 2) an . '3! BI ‘48 r) 1" Ya SC Ii 18 extensive, representative sample of each child's linguistic environ- ment obtained via audio recordings; and 3) once each of these sets of data were described, an attempt was made to determine whether there were systematic patterns of relationship between them. Subjects Due to the necessarily small sample size, several restrictions on subject selection were needed to increase homogeneity of the sample. The subject selection criteria for the study were as follows: only- children of white, intact, middle class, English-speaking families. Mothers' employment outside the home could not exceed ten hours per week, nor could she provide daily child care in her home for other children. Father's occupation and parent education level were not con— trolled. Parent education level ranged from high school diploma to Master's degree, with the mean number of years of education post-high school being 1.74 for mothers and 2.7 for fathers. Father's occupa- tions consisted of the following: 10 were blue collar workers, 6 were white collar or professional, and 3 were college students. A subject pool was first obtained by use of birth records and newspaper birth announcements. A letter of introduction describing the study provided the initial contact with potential subjects (see Appendix A). Each of the families selected received a $30 honorarium in order to provide incentive for continued participation in the study. Given the restrictions, obtaining the proposed 20 subjects proved to be difficult. Due to this factor, and the problem of attrition, the final sample on which the analyses were done consisted of 19 subjects: 10 females and 9 males whose ages ranged from 14 to 17 months at the beginning of the study. The first visit was made when the child had at 19 least three but fewer than 20 words Procedure Part I. Determining Child's Communicative Style. At the beginning of the study the child's current productive vo- cabulary was inventoried by the mother. In addition, the mother was given instructions for keeping a Language Acquisition Diary for all new words acquired by her child (see Appendix B). Words did not need to be phonemically accurate to be considered "true words," but the child's production of the word had to be fairly consistent from one instance to the nextz. While the diaries included imitated words and phrases, as well as spontaneous, only those words which occurred spon- taneously on two or more days were considered part of the first 50- word vocabulary. Words which achieved vocabulary status, and then dropped out of the child's repertoire were still counted in the first 50 words acquired. Each subject was visited every two weeks except when illness or vacations made this interval impossible. The diary information was supplemented by an in-depth interview with the mother at each visit. Interviews reviewed previously acquired vocabulary in order to deter- mine changes in meaning and use of each word over time. For purposes 1Two of the children actually had nearly 30 words at the time of the first home visit. This was due either to inacuracy of parents' esti- mate at the time of the initial phone contact or to a rapid vocabulary development during the short period between the initial phone contact and the first home visit. 2In fact, at this age, there is very little phonemic "consistency" in children's language, as has been pointed out by Ferguson and Farwell (1975). "Consistency" here actually referred to parents' ability to identify the word from one instance to the next. 20 of analysis, each of the 50 words was categorized on the basis of its original use and meaning. After the child reached 50 words, he/she was operationally categorized as "expressive" or "referential," based on the proportion of those 50 words which were general nominals (see below). Once the child had obtained 50 words, the bi-weekly visits were discontinued. Mothers were no longer required to record vocabulary acquisition. However, they were requested to record the child's first word-combinations. When the child had produced ten different word- combinations, 3 final follow-up visit was made. The data collected on word-combinations was exploratory, and will not be considered in the present analysis. Part II. Assessing the Linguistic Environment. When the child had acquired at least 30 but fewer than 503 words, the second part of the study was instituted. In order to characterize the natural linguistic environment of each child, tape recordings were made in the home on two successive days. A voice actuated recorder was used (see Equipment below), with speech sensitive microphones placed throughout the home. It was felt that this extended data collection period would allow more realistic representation of adults' language than would intermittent, short sampling sessions. The rationale here was that while a mother might be able to effectively alter her interaction with the child for short periods of time, it would virtually be impossible to sustain such a 3One child had actually acquired 56 words by the time the taping was done. 21 modification over two consecutive days. In addition, it was assumed that by eliminating the presence of the experimenter, more elements of the "naturalistic" environment could be preserved, as discussed pre- viously. Furthermore, it was assumed that such a procedure would also provide more reliable information regarding the contexts in which adults talk to children (see Friedlander, 1972). Three microphones were placed strategically throughout the home. Ample lengths of connection cord allowed the investigator to keep equipment hidden from view. Microphone wires were placed along floor mouldings, under rugs, and behind furniture. The recording components themselves were placed out of view, such as in a closet, or in an un- used room. The attempt here was to make the recording equipment as unobtrusive as possible in order to minimize families' conscious awareness of its presence. Parents were informed that any speech within a certain range of distance from the micrOphone would activate the recorder. It was ex— plained that, while they were free to turn off the recorder at any time, the purpose of the study is to investigate typical rather than optimal or carefully selected adult-child language interaction, and they were encouraged to leave the recorder on as much as possible when the child was present. Recording equipment was set up the night before the first day's taping. At this time parents were instructed regarding the operation of the recorder. They were instructed that they may turn off the recorder whenever they wished, and were encouraged to turn the re- corder off during the child's naps. They were not to alter their daily routine in any way, with the exception that, in some homes, 22 parents were requested to set the volume of radios and TV's at a lower level. There were a total of three tapes for each child. Tape One was started on Day I as soon as the child awakened, and was left to run until the child retired or the tape ran out, whichever occurred first. At the end of Day I, Tape One was rewound and Tape Two was set up. The experimenter contacted each family at the end of Day I by phone to insure that the tape had been changed and there were no problems. When problems arose or when parents were uncomfortable about making the tape change themselves, the experimenter set up Tape Two. Tape Two was started on Day II as soon as the child awakened, as for Tape One. On Day II, however, the tape was rewound just prior to the child's lunch time. At this time Tape Three was set up and the Akustomat (the automatic start/stop switch) was removed. Tape Three was then left on for the remainder of the day (except for naps, etc., as above). While Tapes One and Two are somewhat compressed because of the operation of the voice-actuated switch, Tape Three represents "real time" and could provide information about amount of child- directed speech per unit of time. Parents filled out an activity sum- mary sheet for each day (see Appendix C) which indicated the hourly activities of the child and parents. Tape One was not included in the linguistic data base for the present study, since Day I was intended to serve as an initial "desen- sitization" period in which parents could become accustomed to the presence of the recorder. Parents were not informed of this, however. All three tapes were audited. The events and activities occurring on each of the tapes were identified and described. 23 Linguistic transcriptions of Tape Two and, in some cases, parts of Tape Three, were done for ten subjects (see below). Transcribers were trained to insure consistency and reliability in transcriptions. (Transcription procedures are presented in Appendix D.) A fifteen minute segment of a pilot tape was used to assess each transcriber's accuracy prior to their transcribing the actual research tapes. All tape segments selected for detailed analysis were re-transcribed by the investigator. Transcription reliability ranged from 84% when utterances of questionable intelligibility were included (for any tape, 10-20% of all utterances were of questionable intelligibility, depending upon the quality of the recording), to 96% when these utter- ances were eliminated. Equipment The following equipment was used in this study: The Uher Model 5000 reel-to-reel tape recorder was selected for its durability and portability, and particularly because of the slow recording speed (15/16 ips) available on this model. This feature together with the use of Scotch 1800 ft. recording tape (five-inch reels) made it pos- sible to record for six hours without changing reels. In addition, use of the Uher, F-411 Akustomat (an automatic start/stop switch which automatically activated the recording system in response to sound, and switched the system off after 5 seconds of silence or low level back- ground noise) allowed maximum efficiency in tape use. A JVC (SEA—30) frequency equalizer was used to enhance the system's frequency response in the range of human speech. This component was added in an attempt to damp out the tremendous amount of ambient noise which is typically present in the home environment. Three highly sensitive 24 AKG-D160 omni-directional microphones, with adjustable desk stands and ample connection wires, enabled strategic microphone placement. The quality and flexibility of this component of the system contributed immeasurably to the system's success. Finally, a Shure M-67 micro- phone mixer allowed separate gain control settings for each micro- phone. To facilitate transcription of the recordings obtained by the system described above, two additional devices were utilized: A Uher F-261 foot control switch eliminated the need for manual operation of the "start", "stop" and "rapid rewind" functions of the recorder. This proved to be a significant time-saving device in transcription and descriptive stages of data analysis. Uher W-214 stethescope head- phones were also employed for easier tape audition. Measures This study consisted of two main variables: Child"s Communicative Style (CCS), based on diary information regarding his/her early ex- pressive vocabulary, and Mother's Communicative Style (MCS) as deter- mined from the in-home recordings. While, as has been acknowledged previously, mother's communication makes up only a part of the child's total linguistic environment, for purposes of the present study, only mothers' language was analyzed in detail. However, since nearly all tapes contain samples of fathers' communication with their child, and since there is very little data available on father language (Berko- Gleason, 1975; Giattino and Hogan, 1975; Friedlander et a1., 1972), this data will be analyzed at a later date. 25 Child's Communicative Style Nelson's categories. Each of the child's first 50 words was cate- gorized using a modification of Nelson's classification system. Only Nelson's major categories, and not her subcategories were used. Each word was categorized as one, and only one, of the following: 1. Specific Nominal: refers to a particular object, person, pet, etc., and does not refer to more than a single exemplar of a class. This includes all proper names. 2. General Nominal: refers to more than one member of a cate- gory of persons, objects, animals, or substances ("ball," "milk," "doggie," etc.). 3. Pronouns: includes words generally corresponding to pro- nouns in adult language ("me," "I," "it," etc.). While pronouns were originally considered general nominals in Nelson's study, they are given a separate category status here, since both Nelson and Bloom have identified use of pronouns as a discriminating feature of two patterns of early language development. 4. Action Words: used to describe, demand, or accompany action ("ride," "byebye," "up," etc.), or to demand attention ("hi," "look," "see," etc.). Unlike Nelson's system, words for social action games such as "patty cake" will be included in this category, as suggested by Benedict (in press). 5. Modifier: expresses attribution ("big, pretty"), state ("hot," "allgone"), location ("there," "outside"), or possession ("mine") . 6. Personal social: expresses an affective state ("want," thank you," "OUCh," "no," I! yes"), or social stereotype ("please, 26 "Okay") . 7. Function Word: performs a grammatical function such as a " "where"), or is used only in relation to other question ("what, words ("for," "is," "to"). Using Nelson's criterion, referential children were those whose SO-word vocabulary contained greater than 50% general nominals; those with less than 50% general nominals were categorized as expressive. Function Categories. Several authors have discussed the impor- tance of considering pragmatic functions in the investigations of early language development (Halliday, 1975; Bates, 1974, 1975, 1976; Dore, 1975; Rodgon, Jankowski and Alenskas, 1977; Wells, 1974). According to Halliday, (1975), the children in the current study would be in "Phase II" of language development. Phase II represents the introduction of the level of form, i.e., most meanings are ex- pressed by means of lexical items which have some recognizable phonemic "form." In Phase I meaning was expressed primarily by intona- tion and gestures. Phase II, which begins at approximately 15 months, is a transition into the adult system of language which consists of three levels: sound, lexicogrammar, and meaning. Halliday reports that in Phase II six different language functions could be identified in the repertoire of his son Nigel: instrumental, regulatory, inter- actional, personal, heuristic, and imaginative. Rodgon, Jankowski and Alenskas (1977) presented a three-dimen- sional coding system for single word utterances: One of these dimen- sions was referred to as "structural linguistic" but actually con- sisted of several categories of language 2E2 as well.4 These included naming, getting attention, demanding object, assertion, habitual 27 association, and so-on. Wells (1976) has studied language meaning and use in two-year-old children. He suggests that any utterance can be categorized three ways: by its "topic" (i.e., its content), "discourse" organization (i.e., how content is organized), and its "interpersonal" features (i.e., its function, such as asking, commanding, etc.) While each of these authors has developed a somewhat different categorization system, they all address the developmental aspects of language BEE in early communication. None of the systems developed by the authors above has been sufficiently well-defined to provide de- tailed analyses of earliest language development. Thus, the current study has empirically evaluated the uses of each child's first 50 words. The following categories of word use emerged for the sample of children investigated: 1. Label: Child names object or person, without making any demand or request. For example, the child notices a ball lying on the floor, points to it, and says "ball," simply to identify the object. 2. Accompanies action: Any utterance which typically accom- panies an action in which the child engages. For example, the child jumps up and down saying "jumping,' or rides a rocking horse and says "rock rock." 3. Response to event: Any utterance which is a response to a particular event. For example, after some one else sneezes, the child says "achoo,' or each time the child hears an airplane flying 4The impossibility of separating language structure and meaning from its function is again apparent here. 28 overhead, the child says "plane." (In this case, plane is not a label because he only uses the word following the event of the plane flying over his house. Were the child to see a plane in some other context, the child would not say "plane.") 4. Response to attribute: Child responds to a particular characteristic of the object such as "hot," "cold," "pretty," "all- gone". 5. Response to question: This category includes responses which the child may use to respond to any yes-no question, for example "yes," "no," "okay," "uhuh" etc. 6. Verbal game: A verbal game consists of any rehearsed re- sponse to a specific question or statement from the parent. For example, the mother says "what's the cow say?" and the child responds "moo ’ H or the mother counts "one, two," and the child says "three." The child does not produce the word in the absence of the specific prompt. 7. Requests: This final category contains several sub-types. The commonality among them is that they all represent attempts by the child to control or manipulate his/her environment. a. Request object: The child uses a word to obtain a de- sired object. The word used may be the name of the object, e.g. "cookie," or a word such as "that" or "want" accompanied by ges- tures (e.g. pointing) which make it clear that the child desires a particular object. This category also includes the words "mine" and "more" since in the current sample these words were always used to obtain something the child desired. b. Request action: The child uses a word to request that 29 some action be performed, e.g. "up" for wanting to be lifted up or "outside" when the child wishes to go outside. c. Request attention: Any word which the child uses primarily to direct attention upon himself is considered a re- quest for attention. For many children words such as "Hi," "see," "Mama" functioned exclusively in this way. d. Request label: The child uses a word such as "that" or "wha'dis" which functions to request the adult to produce a label for an object. Mother's Communicative Style The ways in which one might characterize mothers' communicative style are innumerable. Research on mother-child interaction (e.g. Snow, 1972, 1977; Clarke-Stewart, 1973; Bruner, 1975) has shown that mothers communicate with their children in many different ways, both verbally and non-verbally, and that this communication begins as soon as the child is born. Among the non-verbal factors which have been considered important are eye-contact, joint action (Bruner, 1975), turn taking (Snow, 1977), and immediate responsiveness (Kagan, 1968). Verbal factors which have been investigated can be divided (at least theoretically) into two main categories: the communicative intent or functions of language, and the features of the linguistic code itself. In speech act terminology (Searle, 1969), this might be referred to as the locutionary versus illocutionary force aspects of communication (see also Bates, 1974 and Dore, 1975). The measures to be used in this study to analyze mothers' com- municative style are both linguistic and pragmatic, and have been selected from those reported in the literature which have been found 30 to be related in some way to child language development. 1. Pronominalization. This measure is the ratio of pronouns to nouns which occur in mothers' child-directed utterances. This measure is included because expressive children in Nelson's study were found to use a greater proportion of pronouns in their early vo- cabularies than did referential children. Furthermore, Benedict (1978), who has reported two distinct mother-child interaction styles (referential and action-oriented), found that one of the distinguish- ing features among mothers in a book-reading situation was their noun/ pronoun ratio, i.e. the number of nouns used relative to pronouns. 2. Yes-No Questions. Nearly all investigations of mother— child language have examined the role of questions. Several investi- gators (e.g. Furrow et al., in press) have found that a high rate of yes-no questions in mothers' speech seems to be associated with earlier language development in children, thus suggesting a facilita- tive effect. This measure will be defined as the percentage of child- directed utterances which ask this kind of question, in any form, in- cluding deleted auxiliary ("Ya want some ice cream?"), tag questions ("The doll's going to sleep now, isn't she?"), rising intonation alone ("The dog is in the train?"), as well as the standard subject- auxiliary inversion ("Do you want to go?"). Wh-questions do not fall in this category. 3. Mean Length of Utterance (MLU). This category is included not only because it is probably the most frequent measure used by other investigators in this area, but because there was some sugges- tion in Nelson's study that the language of mothers of expressive children was characterized by somewhat longer MLU than mothers of 31 referential children (Nelson, 1973). 4. Noun Type/Token Ratio. This measure, which is the ratio of number of different nouns to total number of nouns used, is intended to reflect lexical reduction. It has been suggested above that mothers of referential children may make more frequent use of nouns relative to other word categories (e.g. pronouns, verbs, modifiers, etc.). Intuitively, one might also predict that, within the noun category itself, mothers of referential children use fewer different nouns than do mothers of expressive children. This might reflect a mother's being careful, in the early stages of language development, to label an object in one, consistent way, and to name it often. There is some evidence supporting this in Benedict's (1978) study of effects of conversational context on mothers' speech (unpublished data), in which one of the most highly variable characteristics of mothers' speech is noun type/token ratio. 5. Single word vs. Multiword utterances. These measures were designed to be more sensitive than MLU to mothers' use of shorter utterances as a facilitating strategy. The number of one-word utter- ances and the number of utterances which are longer than five words were counted. It was predicted that, since expressive children are more likely to produce multiword utterances as part of their early language, that mothers of these children are more likely to produce long utterances. It was also predicted that mothers of referential children would produce more single word utterances, thus providing these children greater experience with discrete lexical units. 6. Repetitions and Redundancy. Another frequently used measure reported in the literature is one which reflects mothers' use of 32 repetition (Kobashigawa, 1969; Benedict, 1975). Often, repetitions are defined in such a way as to reflect non-compliance on the part of the child. This occurs when repetition of semantic intent without actual repetition of the surface structure or lexical items themselves is considered a repetition. Using this definition the second sentence in the following sentence pair would be considered an utterance repe— tition. a. Get the ball b. I said get it. The purpose of considering various types of repetition in the current study is to determine whether there are differences among mothers in their linguistic strategies for increasing redundancy in their com- munications to the child and whether this is done via the repetition of entire sentences and phrases or via repeatedly identifying the topic noun or action. For this study, in order for an entire utter— ance or part of an utterance to be categorized as a repetition, the surface features must, except for minor deviations, be retained. The following repetition measures were defined for use in the present study: a. Utterance repetition: an exact repetition of an entire utterance which occurs within two utterances following its initial occurrence. b. Partial repetition: any repetition of a.2§££ of an utterance consisting of a word-string of three or more words which appears within two utterances following the original occur- rence of the utterance. c. Noun repetition: Any repetition of a noun which occurs 33 either within the same utterance or within one of the next two utterances. d. Modifier repetition: Any repetition of a modifier de- scribing an attribute such as "big" "dirty" "hot" which occurs either within the same utterance or within one of the next two utterances. e. Action repetition: Any repetition of an action word which occurs within the same utterance or within one of the next two utterances. It is predicted that noun repetitions will be used more often by mothers of referential children, and that utterance and partial repe- titions will occur more freqently in mothers of expressive children. 7. Imitations. Some authors have observed a characteristic of mothers speech which appears to reflect her efforts to maintain a con- versation or dialogue with her child. Snow (1977), for example, has observed the occurrence of vocalization "turntaking" between mothers and their young infants. Halliday (1975) has discussed the importance of the development of dialogue skills in children. One device which mothers appear to use to develop this kind of interaction is that of imitation. That is, mothers imitate the child's utterance, apparently in an attempt to maintain dialogue with a child who has minimal lan- guage skills, and to reinforce his efforts at speech. This measure will simply record the frequency of this phenomenon in the mother's speech. 8. Commands and Prohibitions. Nelson (1975) commented that there seemed to be a tendency for mothers of expressive children to be much more "intrusive,' using more imperatives and fewer questions 34 than mothers of referential children. The measure "commands and pro— hibitions" is intended to evaluate a non-linguistic as well as a lin— guistic aspect of interaction style. All imperative sentences, e.g. "Stop that," "Don't," "Give it to me", as well as use of prohibitions such as "No" or "you can't do that" are counted in this category. 9. Object reference. This measure refers to any reference made by the mother to an identifiable object present in the child's immedi- ate environment. The utterance need not actually contain the object name, but must in some way, comment upon or draw the child's attention to the object. For example, "Go get it," or "Isn't that pretty" are object references just as are the utterances "Go get the ball" and "Isn't that a pretty doll." These measures are designed to assess two levels of mothers' lan- guage. The first five (Pronominalization, Yes-No Questions, Mean Length of Utterance, Noun Type/Token Ratio and Single vs Multi-word Utterances) describe surface structure features of mothers' utter- ances. The last four measures (Repetitions, Imitations. Commands and Prohibitions, and Object References) describe pragmatic aspects of mothers' language and reflect "interaction" or "communication style" in a functional way, i.e. how the mother uses language as she inter- acts with her child. 5It is true that use of surface features alone may miss those utter- ances whose illocutionary force is clearly imperative, while the sentence structure is not. For example, "Will you stop that!" has an imperative meaning, while the structure is interrogative. However, the written transcripts used for this particular analysis did not have sufficient information regarding context and intonation to make judg- ments of semantic intent reliable. 35 Reliability of Measures Reliability scores for the measures defined above are presented in Table 1. Because several of the characteristics being identified occurred infrequently in the mothers' speech samples, the use of a correlatonal procedure, such as a Phi coefficient or product—moment correlation, could significantly inflate the reliability scores. Thus, the following calculation was performed on each measure to determine percentage of agreement. This provided a more stringent measure of reliability than would a correlational procedure. Number of Agreements Number of Agreements + Number of Disagreements. All measures and categorizations were first performed by the investigator. A research assistant was then trained on the coding criteria. Six 50-utterance transcripts of the mother's speech, and three children's first 50 word vocabularies were selected randomly for re-coding by the research assistant. The coding procedures which were used on the mothers speech are included in appendix E. 36 TABLE 1 Coding Reliability for Measures of Mother's Communicative Style and Children's Word Categories ‘ Mother Communication Style Child Communication Style Categories Categories Imitation 84% Nelson Categories (overall) 91% Modifier Repetitions* Use Categories Partial Repetitions 78% (overall) 96% Utterance Repetitions 63% General Nominal 89% Noun Repetitons 82% Specific Nominal 89% Noun Types 86% Action 88% Noun Tokens 91% Modifier 73% Third Person Pronouns 80% Personal Social 100% Commands/Prohibitions 87% Function Word 100% Object Reference 91% Pronoun 100% S-Word Utterances 98% Label 90% Single Word Response to Utterances 91% Attribute 100% Words per Utterance 96% Response to Event 75% Verbal Game 67% Accompanies Action 65% Requests 96% Note: # Agree Reliability = # Agree + # Disagree *Too few instances of modifier repetitions to determine reliability 37 RESULTS The results of this study will be presented in two parts. First, the diary data will be analyzed in order to provide descriptive infor- mation about children's early productive vocabulary, both with regard to Nelson's categories, as well as how children actually BEE their first 50 words. It is on the basis of Nelson's "General Nominal" category that two types of language style--"expressive" and "referen- tia1"--will be operationally defined. The subjects whose first 50 words contain greater than 50% General Nominals will be labeled refer- ential while those subjects with fewer than 50% will be labeled ex- pressive (Nelson, 1973). Second, mothers' communicative styles will be evaluated via results of analyses of the in-home tapes. Mothers of the five most referential and the five most expressive children will be compared with regard to the various measures used to describe mothers' com- munication with their children. Children's Communicative Styles As stated earlier, a total of 19 children were used in this study, ten females and nine males. The average age of the children at the beginning of the study was 15.4 months. The average age at acqui- sition of 50 words was 18.5 months, with a range of 14.8 to 22.9 months.6 The average age at the time of the in—home taping was 17.6 months. 38 Figure 1 presents a summary of all word categories used to des- cribe children's early communicative styles. (These categories were discussed earlier under Method.) While there is some overlap between the two classifications systems (Nelson's and the empirically derived Function categories) there are important differences. Nelson's cate- gories represent a variety of aspects of language, including semantic features (e.g., Function Words), and pragmatic features (e.g., certain sub-types of Action Word category, such as words used to get atten- tion). In order to determine more clearly how children actually BEE their early vocabulary, (in addition to what particular lexical items are included in that vocabulary) the Function Categories summarized in Figure 1 were developed to characterize the communicative uses of the first 50 words of the current sample of children. Table 2 summarizes the mean scores in each word category for all 19 subjects. With regard to Nelson's categories, it can be seen from this table that names for people and things (General Nominals and Specific Nominals), on the average, made up well over half of the first 50 words used by the children in this sample. Action Word was the next largest category, accounting for over 20 percent of the total. The remaining categories of Modifier, Personal Social Expres- sions, Pronouns, and Function Words, were relatively infrequent and 6This finding was approximately one month younger than that found by Nelson (1973) who reported that the average age at 50 words was 19.6 for her sample of 18 subjects. The difference here may be related to the fact that several children in Nelson's sample had siblings. Nelson found that the rate of language development tended to be slower for children with siblings than for only children. This finding is further supported by research by Benedict (1976) whose sample also consisted of only children. Benedict found a mean age which was identical to the one found in the current study of 18.5 months. 39 .mwuos om umpfiu w.awuvHH£o oaauommw ou mom: mmfiuowmumo eyes «0 mcowuwcfimon ou .mH .HOH H muswwm .mvuos uozuo Cu GOHumHmu cw mace pom: ma no snags .umga coaummav as Scam coauuaaw kuwumsamuw w manomuom memo: sowuussm .n mwmawamH vases aw mason many .H .ma Iona cu wafivcoqwmuuou mHHmumamw mvuoa mwvsaocH cacaoum .o hmxo .zox xamsu .mmmoaa .oahuomumuw on .3030 .uamz Hmauom no .mumum 0>Huummmm am mmwwmunxm HmHoom Hmaomumm .m mafia mvfimuso .mpmnu oaomHHm .uo: .aoammommom muumum .wan no coauwooH .muMHm .cowuanfiuuum mmwmmumxm umwwanoz .q oxmu%uumm :.mxmu huuMQ: mm scam mmamw cowuum Hmfiuow you xooH .65: .H: memos mmvsaoafi omH< .cOHucmuum finesse ou no ow .m: .mku cofiuom hcmaeooom no .wcmamv .wnauuwow ou vow: mvuoz cowuo< .m .mmama umaoua Ham mmvaaucH .wmmHu a mo swamamxm cowsm mamcfim m amnu once on ummwu uoa mmov was ..ouo .uo>om .mvwmn .umm .somuwa .uumnno umfisowuuma m ou mumwmm Hmafiaoz camaummm .N ofiwwov .moocmumnam no .mHmsH:m .wuomnno .mcomumn mo .xHHE .HHmn muowoumo m «0 Manama wco swap whoa ow summon Hmaasoz Hmumswo .H m:0fiumuauwmmmfio comaoz mamamxm meoauaaawmo hwow0umo cues 0 l4 .oaw: ou ucwnmn muuomxw was :mfinu: whom was xoon ca Avosafincoov H unsmfim .uommno am now Hanna m nonvonn on ustm wan ummsuon cu mcofiuocsw can onsuoao w on mucnom mango :um:n m.umn3: memos scans wnoB m mom: wafico Hons; umosvom .oH :.¥OOQ: Ho :NSMS: mama was EH: ounuoc .wnomsas wnmsou on nwzuoe mucus vanno coaucwuum w.uH=vm noonfiv on mumamuum wanna conncoun< unmovom .o :.mw: whom mmm .noanomnom on wwumna on on magmas pawnu conuum oEow umzn uwosvon on vnoz a mom: vfifiso sonuo< umwsvom .w =iwena= mama ennnu axon m.an;o mm: mumszmHm :.mnoa: no :manz: mono? any .mnsummw an vowcmnaooom no .mmnsumom an wowcmneooom =umzu: mm zoom vnoB :ncma: no :onooo: wawxmm m .uuwmno can we mama ago on mos new: unos may ha maxooo mumoavon UHHLU .uoonno vmnfimov a :Hmuno on wnos m mom: mango noonno now amonvom .n :.omms .mAMm .ummmMNTUHwaow w on» no wocwmnm can ca vno3 wanna was :szm 300 man can weapons uoc moon mango ace .ucmaoumum no moov nuns: whom nosnoz conumwsv onmwoonm m on uncommon vownmonon >a< meow Hmnnm> .o :.cmoh: mmmm mango .=0Humw=v om%u oclwmz mam cu vuommmn vow conumonv wxwm nosuoz on mom: enaco o:n seas: wmmcoowwn wwvsnocH muonumwso on oncommom .m :.ommw: whom wanna was .mouwwcm mcowaom .ucw>o nwasoaunmm w on mucommon vansu nco>m on uncommom .v |=JAE=||§ .uomnno a mo voow no: mucosou manna onuwanouownmno nmnsonunmm w on wvcoamon vafino muonwnnu< on omcoawom .m 338 soon: 3.8; .mmmmmm an Gnome: Eng monsoon wanvnn onwco cw coauum cm wowsmmEooum cease woswnouu: c4 cowuo< mowcmmsoou< .N .uownno :.HHmn: wzmw com nonmaucwvfl hamfinm vHHnu .ummsvon no menace ham Hams on mucnoa vaano wcnxma usoSnas .cownom no noonpo moan: vanso Hons: .n maofiumonwnmmmHu conuucsm mamsmxm mmmnnacamwo Nwowmumo vnoz Distribution of First 50 Words Among Word Categories for All 19 Subjects Word Category Mean Number Words Falling into Category (Nelson's Classification) General Nominal 24.60 Action Word 11.07 Specific Nominal 6.02 Modifier 3.34 Personal Social 2.86 Function Words 1.12 Pronouns .84 (Function Classification)a Label 26.82 Accompany Action 5.25 Request Object 5.16 Response to Event 4.55 Request for Action 3.75 Response to Attribute 2.75 Request for Attention 2.56 Verbal Game .96 Response to Question .54 Request for Label .43 a Total of Means in Function categories exceeds 50 because some words were classified as having more than one function. 42 made up the remaining 16 percent of the first 50 words. It can also be seen from this table that the most frequent use of the first 50 words was to label, with an average of approximately 27 words being used in this way. An average of 12 of the first 50 words were used to make various requests or demands, and 5 were used to accompany actions. The remaining words were used as fairly stereo- typic responses to events, yes-no questions and well-rehearsed verbal games. As has been pointed out by Nelson, these kinds of classifications of the nature of each of the child's first 50 words does not charac- terize the frequency with which the child used a given word. For example, while a child's repertoire may contain some 30 or 40 words, it would not be surprising that 80 percent of that child's language output for any given day consisted of a single, potent word such as "Mama" or "look." While these categorizations give us some sense of the character- istics of children's early vocabulary, in general, the more important question for the current study is how do children differ in the de- ve10pment of early language. The next section addresses this issue and and summarizes the results of comparisons between two groups of children who differ somewhat in lexical deve10pment and use of lan— guage. Group7Comparisons Table 2 presents the average number of the first 50 vocabulary words in each of the Nelson categories for the entire group. The number of these first 50 words which were general nominals ranged from 10 to 38 for the 19 subjects. Using the criterion described earlier, 43 those children who had more than 25 General Nominals in their early vocabulary were labeled expressive. While the average age at acquisi- tion of 50 words was greater for expressive than for referential children (E = 19.16 months as compared to 17.67 months), this differ- ence was not found to be statistically significant (t(l7) = .93, p :> .05). Of the 19 subjects, seven were referential and twelve were ex- pressive. It should be noted that, of the seven referential children, the parents of all but one were college educated. The mean number of years of post-highschool education for parents of referential children was 3.14 and 6.72 for mothers and fathers respectively. For parents of expressive children, the number of post-highschool years of educa- tion was .91 for mothers and 1.25 for fathers. Product moment corre- lations between the number of General Nominals in the child's first 50 words, and number of years of parents' post-highschool education were computed for mothers and fathers separately. The results of these correlations suggest a significant positive relationship between edu- cation level and use of General Nominals (r = .58 and .66 for mothers and fathers respectively). This finding is an important one and will be considered further in the discussion. Nelson's Categories. Table 3 presents the mean number of words in each of Nelson's categories for the two groups. Significant dif- ferences were found between the two groups in the categories of Action Words (t(l7) = 2.91, p <1 .01), Personal Social Expressions (t(l7) = 3.26, p < .01), Pronouns (t(l7) = 2.30, p < .05), and Function Words (t(l7) = 2.18, p <1 .05). Referring again to Table 3a, it can also be seen that the mean 44 TABLE 3a Differences Between Expressive and Referential Groups in Number of Words in Each Nelson Category Referential Expressive Proportion Proportion Mean of first Mean of first t Category (n87) 50 words (n-12) 50 words value General Nominal 30.6 .61 18.6 .37 ----5 Specific Nominal 5.14 .10 6.91 .14 1.46 Action Word 8.85 .18 13.3 .27 2.90** Modifier 2.86 .06 3.83 .08 1.76 Personal Social 1.43 .04 4.30 .08 3.26** Pronouns .42 .01 1.25 .02 2.30* Function Words .57 .01 1.66 .03 2.18* Word-Combinations .57 ____a 2.50 ____a 2.57* aNot applicable to word-combinations bSince the number of subjects in the two groups was unequal (referen- tial n-7, expressive n-12) it was necessary to compute t' for those comparisons in which the variances were not homogeneous. Variance homogeneity was determined using a simple test proposed by Hartley as follows: (see Winer, 1971, p. 206) 2 95 (2,11) = s F max. largest treatment variance 32 smallest treatment variance Because t will either underestimate or overestimate the true t value, depending upon whether the (footnote b continued on next page) *t(df-17) < .05 **t(df=17) < .01 45 number of word-combinations which occurred prior to the acquisition of 50 words was much greater for expressive than for referential children (2.50 as compared to .57). Inspection of the raw data suggests that there is a relationship between the use of word-combinations and the categories of Pronoun and Function Word. Of all word-combinations produced prior to acquisition of 50 words, 65% contained either Function Words or Pronouns, or both. Figure 2 presents examples of these types of utterances. Thus, the greater incidence of early word- combinations among expressive group children is at least partially responsible for the greater proportion of Pronouns and Function words. There is evidence that the term ' 'word—combinations" is inappro- priate here. The term "multiword string" will be used to refer to word combinations occurring prior to the acquisition of 50 words. There is little evidence that these early combinations are true b’ Table 3a (continued) larger sample has the larger or the smaller variance, t' was computed for those categories where variances were found to be unequal. The following computation of t' was performed formed according to Snedecor and Cochran, 1967, 114-116): - - 2 2 ' = — t (x1 x2)/ \/ (s1 /..1 + 32/112) The significance level of t' is approximately: 2 2 (wlt1 + w2t2)/(wl + w2), where wl - 31 /n1, w2 — 32 /n2 Unequal variances were found for three categories: Specific Nominal, Verbal Game, and Request for Object. t' was computed for each of these, and the results are presented in Table 4. None of these values reached the .05 level of significance. 46 What this? ‘ng dog. lhgt t_ha_t? t wanna. Me too. I love ypp, Where Papa? Here y'are. I eat. Help mg. .My Dad. ‘Whgtg go? .12 _i_? H2222 .12 .i_t_? Figure 2 Examples of Multi-word Strings Containing Pronouns or Function Word syntactic combinations of lexical items. They were essentially used as one-word utterances; and their components were almost never used alone or in combinations with other words. In summary, then, it can be said that referential children's early vocabularies contain mostly names of things (general nominals), some action words and names of people (specific nominals), occasional modifiers and social expressions, but almost no pronouns, function words, or multi-word strings. Expressive children, on the other hand, were likely to have nearly as many action words as General Nominals, in their first 50 words. Although they too used Pronouns, Function Words and Personal Social expressions relatively infrequently, the used these words twice as often as did referential children. They were also more likely to produce multi-word strings prior to the 47 acquisition of 50 words than were referential children. Finally, no single category predominated for expressive children to the extent that General Nominals did for referential children. Function Categories. As discussed earlier, a categorization system of language use was empirically derived using the diary data. Table 3b presents the mean number of words in each of these function categories. Unlike the Nelson categories, it was possible for a word to be categorized in more than one way if a mother reported that the child was as likely to use it one way as another. Interestingly, new vocabulary items seldom had more than one function. Only as the word became well-established as part of the child's language repertoire did it take on new functions. Halliday (1975) also observed this tendency in Nigel's early language development. The greatest difference between the two groups was in the use of language for labeling. The difference between a mean of 33 for the referential group and 20 for the expressive group was highly signifi- cant (t(17) = 6.51, p < .01). This finding is, of course, not sur- prising given the definition of the two groups. Most general nominals are used as labels. However, it should be pointed out that the cate- gory "Label" used here is not_§ynonymous with Nelson's "General Nominal" category. In Nelson's scheme, General Nominal refers to the meaning, rather than the use of a word. Thus a word which was used only to request an item, e.g. "bottle", would be categorized the same as pointing to a toy and saying "ball." Both are names of things and therefore categorized as General Nominal. In the use categorization system, however, the words "bottle" and "ball" would be categorized differently: while "ball" would be considered a label, "bottle" would 48 TABLE 3b Differences Between Expressive and Referential Groups in Number of Words in Each Function Category Referential Expressive Mean Mean Category (n-7) (n-12) t value Label 33.14 20.50 6.51** Accompany Action 3.85 6.66 2.62* Response to Attribute 2.00 3.50 2.80* Response to Event 3.85 5.25 1.27 Response to Question .43 .66 .62 Verbal Game .85 1.08 .89a Request for Object 4.42 5.91 .83 Request for Action 3.10 4.41 1.18 Request for Label .28 .58 1.76a Request for Attention 1.71 3.41 2.17* 8Value of t'. See Footnote b, Table 3a *t(df=17) <: .05 **t(df—l7) <: .01 49 be categorized as a Request for Object. A significant difference was also found for the category referred to as "Response to Attribute." Responses in this category were mostly "modifiers" which described a particular feature of an object or experience, such as "soft," "pretty," "hot," "dirty," "cold," "etc. An average of 3.5 words were used in this way by expressive children, and 2 by referential (t(l7) = 2.80, p ‘<:.05). Inspection of the raw data reveals some interesting differences between the two groups. A major contribution to this category for expressive children was made by negative descriptors such as "dirty," "yuk," "stink," etc. How- ever, these words were almost non-existent among referential children. (The only occurrence of this type of word use in the repertoires of referential children was one child's use of the word "dirty.") Another category in which a significant difference in language function was found was the category "Requests for Attention." This category included all utterances whose primary purpose was to focus the caretaker's attention directly upon the child. These included such words as "mom," "Hey,", and "Hi." Expressive children used these words significantly more often with a mean of 3.41 out of the first 50 words, as compared with a mean of 1.71 for referential children (t(l7) = 2.17, p < .05). Three other categories of language function developed in this study are similar to the Request for Attention category in that they reflect children's efforts to control their environment: Request for Object, Request for Action, and Request for Label. While the mean in each of the four Request categories is greater for the expressive groups than for the referential group (see Table 3b), the differences 50 between the two groups reach significance only in the Request for Attention category. When the data are grouped together under a single category called "Requests", the difference between the group means does approach significance (t(l7) = 1.71, p ‘<:.10). Inspection of the raw data reveals another difference between the two groups related to use of language to control one's environment. This was found in the frequency with which the words "more" and "mine" were used. For the children in this sample, the function of these words is fairly consistent. "More" is a request for more food, and "mine" is an emotional request for an object in someone else's posses- sion. These, together with the word "mama" or "daddy" used to request the presence or attention of a parent or caretaker, are clearly the most effective control words in a child's repertoire. Two referential children (28%) used the word "more", while six of the expressive children (50%) used this word. An even greater difference existed for the use of the word "mine." Only one referential child used this word (14%), while seven of the expressive children (58%) used it. No referential child used 22th "more" and "mine" while five of the expressive children used both. While it is not reasonable to subject these differences to statistical analysis due to the small and unequal n's, these findings are, nevertheless, suggestive of differences between the two groups. The final category of language function which revealed signifi- cant differences between the two groups was the category referred to as Accompanies Action. This category includes words used to accom- pany the child's own action. For example, the child says "oh oh" while falling, or says "jump" while jumping, or "beep beep" while 51 pushing a toy car. As can be seen in Table 4, the mean for the refer- ential group in this category was 3.86. For the expressive group the mean was a significantly greater 5.25 (t(l7) = 2.62, p < .05). In summary, then, it appears that, while referential children are much more likely than expressive children to use language primarily to label things in their environment, they are less likely than expres- sive children to use language to request attention, to accompany actions, or to respond to attributes, particularly negative attri- butes. There is some suggestion that, in general, expressive children make greater use of language to manipulate and control the behavior of people in their environment than do referential children. This is demonstrated by greater use of words like "more" and "mine" used as demands. Mother's Communicative Styles As described earlier, mothers' communicative styles were observed via in-home tape recordings. Using the procedure described earlier, a total of 57 tapes should have been produced (three for each of the 19 subjects). However, due to occasional human error, or mechanical mal- function of the recording system, two tapes were blank, and four were of relatively poor quality. Thus a total of 51 usable tapes was pro- duced. Since these tapes ranged from approximately three to six hours in length, producing approximately 250 hours of taped material, it was important to devise some sort of data organization procedure which would allow efficient access to the data. The time required to make actual verbatim transcription of the tapes was approximately four to five hours for every hour of tape. Thus, transcription of all 51 tapes was not feasible. Each of the tapes was, however, audited and 52 described in its entirety. From these auditions, a catalogue of the contents of each was made. The catalogue recorded the activities and interactions which occurred within sequential intervals of 30 recorder counter units. On a separate sheet, the location, by tape number, of several target events (e.g., breakfast, play, book reading, etc.) was separately recorded. Using this separate sheet, supplemented by the catalogue, all key interactions for each subject could be reliably located for analysis. In an effort to minimize the time required for data reduction, and to maximize any differences which might exist between the mothers of the two groups of children, tapes of the five most referential and the five most expressive children were selected for analysis of mothers' communicative styles. Nelson's General Nominal category was again used to define the two groups, with those children having the most General Nominals comprising the referential group, and the five children with the fewest making up the expressive group. Once the ten mothers to be used in the analysis of mothers' com- municative styles were identified, a procedure had to be devised which would allow comparisons between the two groups which would be uncontaminated by situational effects. It was thus necessary to find situations which could be operationally defined and which occurred on the second day of taping for each of the ten mothers.7 It was predicted that two situations which might meet these criteria were a breakfast segment, and mother-child play segment. "Breakfast" was operationally defined as the 50 consecutive child-directed utterances which occurred following the mothers first mention of breakfast.8 The number fifty was selected because all breakfast segments contained 53 fifty or more such utterances. In other words, some segments con- tained no more than fifty appropriate utterances, making fifty the maximum sample size which could be used. The operational definition of "Play" proved to be somewhat more problematic. Prior to the execution of the study, Play had been de- fined as a playful interaction between mother and child which a) in- volved some use of materials, such as toys, etc., b) was not a book reading situation, and c) was one in which the mother's attention was focused exclusively upon the child. It was discovered, however, that any type of mother-child play was a relatively low frequency occur- rence for nearly all of the mother-child pairs. Furthermore, play segments of the type described above almost never occurred in some children's homes. This finding becomes even more interesting when the expressive and referential groups are compared with regard to amount of time spent playing with their children. While it is not practical to determine precisely the amounts of time each of the 19 7As stated earlier under Method, the first day of taping (tape one) was not used in the analysis of mother's communicative styles, since it was believed that the effects of the presence of the recording system on mothers' speech would be smaller on the second day (tapes two and three). It was the subjective impression of the experimenter that, in fact, only the first few minutes of tape one were influenced significantly by the presence of the recording system. The nature of the family interactions and communications on the remainder of the tapes was often personal and intimate, suggesting that subjects were reasonably natural in their interaction. In interviews following the taping, several parents reported that they tried to be aware of the recorder's presence, but were unable to. This is probably due to the fact that, in most cases, most of the recording equipment was not visible. 81m one tape a totally different activity-dressing-followed the mother's first mention of breakfast. In this case, the fifty utter- ances following her next mention of breakfast were used which was when the breakfast situation actually took place. 9+ mothers spent playing with her child, it was possible to count the number of 30-unit recorder intervals9 which contained mother—child play, and thus provide an estimate of the frequency with which this type of activity occurred. Table 4 contains the average number of occurrences for each group for tapes one and two. (Tape three could not be compared due to missing or poor quality tapes.) Any playful interaction with the child in which the mother's attention was focused exclusively upon the child was counted as play. Book reading segments were also considered a type of play, but were tallied separately. These numbers suggest substantial differences between the two groups of mothers in the amount of time spent in these kinds of activities. TABLE 4 Average Number of 30-Unit Recorder Intervals in which Play or Book Reading occurred PLAY READING Tape 1 Tape 2 Tape 1 Tape 2 REFERENTIAL GROUP MOTHERS 9.4 4.6 3.2 3.6 EXPRESSIVE GROUP MOTHERS 2.8 .8 .O .O Thus, in order to find play segments for the five expressive group mothers, it was necessary to redefine Play to include any kind 9It must be pointed out that the actual time required for a 30 counter unit interval on the recorder increases from the beginning to the end of the tape, with an approximate range of one-and-a-half to four minutes. The assumption here, of course, is that there is no differ- ence between expressive and referential groups with regard to whether play segments occur at the beginning or end of a tape. This assump- tion has not been tested in the current study. At best, this measure offers only a crude estimate of time spent in mother-child play. 55 of playful interaction such as tickling, teasing, etc. Furthermore, in some situations it was not possible to find a play segment long enough to provide a fifty-utterance sample. When this was the case the remaining number of utterances which were needed to make a total of fifty were obtained from the next play segment which occurred on the tape. Nearly all segments for both Breakfast and Play were ob- tained from tape two. However, in two cases it was necessary to obtain the Play segment from tape three. The actual data analyses for this part of the study were carried out on two 50 utterance samples for each of the ten subjects (five mothers of referential, and five of expressive children). A total of 20 measures (summarized in Figure 3) were performed on each 50-utter- ance sample. A 2 x 2 two-way analysis of variance (group by situa- tion) with subjects repeated on the situation factor, was the statis- tical analysis used to evaluate differences between the communicative styles of the two groups of mothers, as well as the effects of situa- tion upon communication, for each of the 20 measures. Table 5 summarizes the results of the analysis of mothers' com- municative styles. For most of the measures, the numbers on this table refer to the mean number of occurrences of the particular feature being measured which occurred in a 50-utterance sample. Be- cause both the subject sample size and the linguistic sample size are very small, it was determined, a priori, that a significance level of .10 or better would be considered meaningful for the present study. As Winer (1971, p. 14) points out, "The frequent use of the .05 and .01 levels of significance is a matter of a convention having little scientific or logical basis." He goes on to state that higher levels 56 may be appropriate, particularly when power of tests is low under lower levels of significance and when type I and type II errors are of equal importance. While the data from the current study do provide information regarding the nature of mother-child communication in general, this topic has been frequently discussed by other authors (See Snow, 1977 for review of this literature) and will not be considered in the present study. The focus of the current study is upon the differ- .EEEEE in mother's speech to their children. These data are summar- ized in Table 5. Group Comparisons Table 5 summarizes the means for the two groups of mothers, for each measure, in two situations, Breakfast and Play. Each of these measures will be discussed below with regard to the differences between mothers of expressive and referential children. Noun Pronoun Use. Examination of Table 5 indicates that while there is no support for the prediction that expressive group mothers would make greater use of Third Person Pronouns (F(l,8) = .653, p > .10) or Total Pronouns (F(1,8) = 1.288, p >.10) than would mothers of referential children, there is fairly consistent support for the notion that mothers of referential children do use more nouns than do mothers of expressive children. Mothers of referential children use more Noun Tokens (F(1,8), = 4.164, p_= .07), than mothers of expressive children. Furthermore, when Proper Nouns (names of people, etc.) are eliminated from the noun count leaving only Common Nouns, the differences are even greater for the two groups (F(1,8) = 4.017, p = .08, and F(1,8) = 6.867, p_= .03 for Common Noun Types and l 57 moamum m>nuwoncsaaou .mnoanoz mumSHm>o On mom: monsmmoe mo unmasom m onownm .Ao>onm q .0: momv maoxou csoc cosaou on moazu anon aoEEoo mo oHnmn age .oHaswm oocmnmuuzlom may ca vow: mason coasoo «0 names: Hmuou ofie .onemm oucmnwuHSIom onu cw mason coEEou udonmwwam mo nonaoa can monsmwma was .o>ocm ucsoo cso: mzu Eonu mason nmmona HHm mmumcqaaam mnswmme wane .anHme moEoomn canon man .mowmmno:a mason ou m>nuwamn .mcsoconm mo noses: any mm .msse .mcsocona Hmuou + wcmeu csoc\mamxou no: mo wouaaaoo xfiamsuom ma mane .oaaemm mocmnwuuslom mSu canuws wafinnsooo masoconm mo nonasa HmuoH .mnfimcu .nnmnu .wnH .mnm: .no: .mn: .ommcu .owonu .swzn .no: .Ew: .zmzu .uH .msw .o: “A man .3ocm ma wmcfiwmw mmv w=H30HHow man ovsaocfi masoconm :omnon unwna .manemm mocmanHSIom m an wannnsooo masocona comnom unfinu mo noses: HmuOu och .hucmvcovon anon may mH nwuwwnw any .mH owumn mafia noHHmEm osu .w=£H .ooao maco mom: was coo: unoanMHw comm umzu oumonwcfi vasos Songs o.H hamsofi>no ma oaumn :mxou\om%u answxma one .mHaamm moswnouuslom m now mcoxou coo: Houou on momma soon «0 noses: Hmuou mo canon ace .mamxou coon OH mam mahu csoa oao mucommnmmn mwnu .mmafiu OH mom: mw coo: mamm m5» ma .msne .mHQEmm any a« vow: mono: mo nonesc kuou may .mamemm mocmnmuuslom ca mom: mason acmnmmmav mo nonsss HmuOu one .maasmm mocmnmuus on a now mocmnmuun m.no£uos non mvnoz mo noses: owmnm>m onn monsmmoa .qu am On wonnowon coumo .mnawmoe mack owumm coxow \mahe asoz aoeeoo mcoxoa adoz :oaSoo amaze csoz coEEoo onumm asoconm\asoz masoconm Hmuoe masoconm cownom vnwne oaumm :oxOH \mmms csoz mcoxoe csoz momma csoz mocmnmuub mo nuwamg can: connnannma onswmmz 58 Awosnfiunoov m onswwm .onaom oononouuolom onu an nsuoo nonns vnos onwch o no wanumnmcoo mooconouus mo nonaaa Houou one .oamaom oonononus Ion onu an nsooo nonna onoe no mvnos o>wm mo moononoun: mo nonasc Houou one Azmsoono oon oEom news 30% on: .m.ov conmno>cn enonnwxsoluoofinsm unaccouw onu woo A:~ow so»: ..w.ov onoHo nonuosonan wcwwnn .Azesom u.:ov .Eoono ooH oEom ucos now: ..w.ov .mconnmosc won A:waoono oon oEom woos we: .w.ov women enonnwxso mononov ovsnonn omone .onaeom oocmnouu: on onu an nsooo noHnB ouonnmonnno on vasos :oa: no zoo»: mo omcommon o nonns on .snow eno an .m:0fiuwo=v mo nonasc one .vonnauoo umnam nonmnvoe onu nonns an ooconouua onu wcnsonaom mooconouua can axon onu mo oco cw no oononouu: oeow onu annuwz nonuno .uomolonlmH nonmavoa o moanu mo nonaoa one .wonnsooo umnww wnos conuoo onu nonns aw oocononu: onu wanoHHow mooconouus oau uxo: onu mo oco an no oosonouu: oaom onu Cannes nonuwo .vomsnon on wnos nowuoo co moan» mo nonasc one .wonnoooo umnnw noon on» nunns an ooconouus onn wcwaoanow mooconouu: o3» uxoc onu mo one an no oonononu: oeom onu nwnuws nonnno .cnowo mom: on anon o mosHu mo none:n one .onnaom ooconouuolom onn an nsooo nonna Aooconouua onu mo oononnoooo Honnwnno onu monsoHHON woocononu: osu cannns mnooano nofins .mvnoB onoa no oonnu mo oonoovow o «o wcnumnmcoov oononouua no mo unom o mo mconunnomon mo nonasc one .onoaom oononouualom onu an .oononnsooo Honuwan mun wcn3oHHom wooaonoun: oau :nnuH3 nnooo nonna .ooconouus onnuco no mo moonuwuoaon nooxo mo nonasa one wooco Inonns wnoB onnfim mooconoun: wnon mnowumoso oz\moe conunnmaom nonnneoz conunuoaom conuu< conufiuomom naoz mconnfinomom nonunom moonnfiuomom oononouua .wn .en .cn .mn .qH .mH .Nn .nn cowuwcfiwon onSmooz 59 Amosnnncoov m onswnm :.nsov an non: ..w.o .uH manao: annosnoo usonnHB noohno onn on nowon hos onm .uaosnonn>co ononvoaan m.vnwno onn an naowona on nonnB no8nao no .comnoq .uoonno no on mnowon nonnoa onu moann no nonasc one moononomom noonno .nN .onaaom oonononuSIom onn an noooo nonns A:.nonn on n.aoo soe: no :02: muonnnnnnonm .w.ov mconnnnnnona no A:.naow now: ..w.ov mooconcom o>nnonoman «0 none:: one a ovaoaaoo .oN .onsom oonmnonuslom onn an wanna onn en owoa oocononns so monounsn nonnos onn moann no nonasc one mconuonnaH .an nonnnannoa onsmooz 60 8. v 9...... mo. v not 3. v as cue. aoo.n «eno.¢ An.¢v ¢.e w.on A~.oHv ~.m ~.nn momee ccoz coaaoo «mm. «mom.m «em. A¢¢.v no. we. Anm.v om. Nm. canon ccoconm\ccoz woo. an. ww~.n Ao.omV o.mm «.mm An.omV c.o~ o.mm occoconm Honoe new.~ mno. mmc. ~.o N.¢ N.m o.m o.m e.q mccoconm comnom wnfine qnn. mnw.n «moo.¢ Amm.v mm. we. Ace.v ea. no. canon coxoe \ocme ccoz mmo. aammm.m «acn.¢ An.mmv o.wn o.e~ Am.nmv ~.o~ q.om mconoe ccoz com. «aomo.o «mm.~ Am.nnv o.¢ o.m~ Am.mnv ~.~n w.¢n momee ccoz coo. mwm.n NNH.N An.mV m.~ c.m A¢.mV N.m e.m oocononna no nnwcon coon m m m vocnnaou coo: coo: eocnnaoo coo: coo: mmmbm