MSU RETURNING MATERIALS: Place in book drop to LIBRARIES remove this checkout from w your record. FINES um be charged if book is returned after the date stamped beIow. SUBCATEGORIES OF ACTION VERBS IN CHILDREN’S LANGUAGE By Sharon Lynn McWhirter A THESIS Submitted to Michigan State university in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Audiology and Speech Sciences 1988 ABSTRACT SUBCATEGORIES OF ACTION VERBS IN CHILDREN’S LANGUAGE By Sharon Lynn McWhirter Semantic relational features of children’s utterances have been described using global categories which do not adequately describe later language development or reveal differences between language- impaired and normally developing children. The purpose of this study was to apply subcategories of action verbs to the spontaneous language samples of eight nonclinical children and to determine if these subcategories could distinguish the action verbs of two normal and two clinical children matched on the basis of age, sex, and socio-economic status. The results revealed four trends in verb distribution: 1) Movement verbs were more frequent than Nonmovement verbs; 2) Change of State verbs were more frequent than Nonchange of State verbs; 3) among Movement verbs, from.most to least frequently occurring, were Change of Locative, Attributive, and Pessessive States, respectively; and 4) among NOnmovement verbs, no differences existed.among NOnchange of Pesitional, External, and.POssessive States. Significant differences were not found between the normal and language-impaired children. ACKNOWLEDGMENTS This project could not have been ccmpleted.without the assistance of several people to whom I would like to express my sincere gratitude. First, I would like to thank Dr. Ida Stockman, Chairman of my thesis committee, for her insight and guidance throughout this investigation. I would also like to express my appreciation to the members of my thesis committee, Dr. Gloriajean wallace and Dr. Philip Davidson. In addition, I am indebted to those children who participated as subjects in this study. Finally, I would like to thank my family and friends for their patience, love, and support throughout this endeavor. This work was made possible through research funding from both a National Science Foundation grant (1BN8841—8587) and a Michigan State university All-University Initiation Research Grant, Dr. Ida Stockman, co-principal and principal investigator, respectively. iii TABLE OF CONTENTS LIST OF TABLES LIST OF FIGURES INTRODUCTION Statement of the PrOblem The Use of Gldbal Semantic Categories in the Description of Later Language Development The Use of Global Semantic Categories in the Differentiation of Normal and Language-Impaired Children Definition of Action Justification for Looking at Subcategories of Action Summary Purpose REVIEW OF THE LITERATURE subcategories Based on the Adult Language System subcategories Based on Nonmal Language Development Summary PROCEDURES Part 1 -- Action subcategories in a NOrmative Sample iv Page vii 16 17 22 24 25 25 29 34 36 36 v Description of the Data Base Selection of Data for Study of Action Subcategories Treatment of the Data for Analysis Procedures for Assigning Subcategories of Action Utterances Observer Agreement in the Subcategorization of Action Compilation of Results Outcomes Which Would Support the Ekistence of Subcategories of Action Utterances in Normally Developing Children Part 2 - Action Subcategories in Clinical and Normal Samples Description of the Subject Characteristics Subject Selection Procedures Procedures Used To Obtain language Samples Form of the Data Tabulation of the Data Outcomes Which Would Support the Argunent of Differences in Semantic Relations Used by Norml and Clinical Groups RESULTS Part 1 - Action Subcategories in a Normative Sample The Relative Frequency Distribution of Verb Subcategories The Diversity of Verb Poms Within Each Subcategory Part 2 - Action Subcategories in Clinical and Normal Samples The Relative Frequency Distribution of Verb 37 39 Page 39 42 49 53 53 54 54 59 61 63 64 64 66 66 67 74 86 vi Subcategories for Normal and Language-Impaired Groups 82 Page The Diversity of Verb Forms Within Each Subcategory 89 DISCUSSION 96 Summary and Conclusions 96 Comparison of the Results With Other Studies 98 Interpretatioin of the Distribution of Action Verbs 100 Interpretation of the Results of the Comparison of Normal and Language-Impaired Children 107 Implications for Future Research 110 APPENDICES 113 LIST OF REFERENCES 147 Table 10 11 12 LIST OF TABLES Results of Inter-judge Reliability Measures. subject Characteristics. The Percentage of Different Action Verbs In Each Category for Each Child in Normative Sample The Percentage of Different Action Verbs In the Movement and Nonmovement Categories The Percentage of Different Action Verbs In the Change of State and Nonchange of State Categories Irrespective of Movement or Nonmovement Classification. The Percentage of Different Movement Action Verbs In the Change of State and Nonchange of State Categories. The Percentage of Different Movement Action Verbs In the Change of State and Nonchange of State Categories. The Percentage of Different Movement Action Verbs In the Change of Locative, Attributive and Pessessive State Categories. The Percentage of Different Nonmovement Action verbs In the Nonchange of Pesitional, Internal and Pessessive State Categories. The Diversity of Action verbs In the Movement- of Movement—Change of Locative State Category. The Diversity of Action Verbs In the Movement- Change of Attributive State Category. The Diversity of Action verbs In the Movement— Change of Possessive State Category. vii Page 52 58 68 70 71 71 72 73 74 75 76 77 Table 13 14 15 16 17 18 19 20 21 22 23 24 25 viii The Diversity of Action Verbs In the Movement- Nonchange of State Category. The Diversity of Action Verbs In the Nonmovement- Change of State Category. The Diversity of Action Verbs In the Nonmovement- Nonchange of Pesitional State Category. The Diversity of Action Verbs In the Nonmovement— Nonchange of External State Category. The Diversity of Action Verbs In the Nonmovement— Nonchange of Possessive State Category. A Comparison of the Percentage of Different Action verbs Within Each Category Between Normal and Language-Impaired subjects. The Percentage of Different Action Verbs In the Movement and Nonmovement Categories for Normal and Language-Impaired Subjects. The Percentage of Different Action Verbs In the Change of State and Nonchange of State Categories for Normal and Language-Impaired subjects, Irrespective of Movement or NOnmovement Classification. The Percentage of Different Movement Action Verbs In the Change of State and Nonchange of State Categories for Normal and Language-Impaired subjects. The Percentage of Different Nonmovemnt Action verbs In the Change of State and Nonchange of State Categories for NOrmal and Language-Impaired subjects. The Percentage of Different Movement Action verbs In the Change of Locative, Attributive, and Possessive State Categories for Nbrmal and Language- Impaired subjects. The Percentage of Different NOnmovement Action.Verbs In the Nonchange of Poaitional, External, and Possessive State Categories for Normal and Language- Impaired subjects. The Diversity of Action Verbs In the Movement—Change of Locative State Category. Page 78 79 79 80 8O 83 86 86 87 87 88 89 90 ix Table 26 The Diversity of Action verbs In the Movement- Change of Attributive State Category. 27 The Diversity of Action verbs In the Movement- Change of Attributive State Category. 28 The Diversity of Action Verbs In the Movement- Nonchange of State Category. 29 The Diversity of Action verbs In the Nonmovement- Change of State Category. 30 The Diversity of Action Verbs In the Nonmovement- NOnchange of Positional State Category. 31 The Diversity of Action Verbs In the Nonmovement- Nonchange of External State Category. 32 The Diversity of Action Verbs In the Nonmovement- Nonchange of Possessive State Category. Page 91 92 93 93 94 94 95 Figure LIST OF FIGURES Subcategories of Action Based Upon the Interaction of Movement and Change. The Mean Percentage of Different Action verbs 1_1 Standard.Deviation Within Each Category.' A Comparison of the Mean Percentage of Different Action verbs Within Each Category, Between Normal and Language-Impaired Children. Page 48 69 84 INTRODUCTION During the past two decades, research on early linguistic development shifted its focus from.the syntactic structures the child produces to the meaningful relationships expressed by word combinations. Prior to this shift, empirical studies of the 19603 gave rise to syntactic descriptions of language which were often based on Chomsky’s (1965) transformational grammar. Using this model, highly structural descriptions of language were provided by analyzing sentences according to their deep underlying structures and transformational rules. As a result, sentences were often described according to their constituent parts. For example, the sentence, "the boy is riding a bike," would be described as "Noun Phrase + verb Phrase," or more specifically in terms of constituent structure of noun phrase and verb phrase, "Article + NOun + Auxiliary + verb + Article + NOun." Despite the valuable insight this model provided into the syntax of children’s language disorders, there was little consideration given to the meanings underlying these syntactic structures. [A theoretical shift in thinking toward a semantic perspective occurred.during the 19708. Bloom.(1970) was one of the first investigators to systematically describe the semantic relational properties of children’s utterances. She noted.that utterances using the same syntactic properties were produced.by children to 2 express different semantic relationships. Bloom’s method involved "rich interpretation" of utterance meaning, derived from examining the utterance’s structural characteristics in relation to its nonlinguistic context. The nonlinguistic context referred to the events occurring at the time of the utterance. For example, if a child said "gimme juice," the nonlinguistic context might consist of the child.pointing toward a glass of juice in his mother’s hand. Thus, semantic relationships extended.beyond lexical referential meaning to include meanings expressed by relations between words (Leonard, Bolders & Miller, 1976). For example, Bloom.argued that when a child says "mommy sock," his expression may convey something about the relationship occurring between an agent (mommy) and an object (soCk), if the situational context of the utterance was that of his mother putting on the sock. The same syntactic utterance could also be used to express the relationship between a possessor (mommy) and a possession (sock). The relational meaning was generated by the combination of at least two words and resulted in a compositional meaning that connoted much more than the lexical meanings of the words alone. 0n the basis of her research, Bloom (1975) was able to identify the following semantic relations of two—word sentences: 1. Existence 2. NOnexistence 3. Recurrence 4. Agent, Action, Object 5. Passessive He (is 6. Attributive 7. Locative In addition to these semantic relations, Bloom, Lightbown, and Ecod (1975) also identified the following categories as emerging earlier in children’s language development: 1. Locative Action 2. Locative State 3. NOtice Following Bloom, Lightbown, and Reed’s (1975) research, a variety of systems were developed for categorizing semantic relations. Schlesinger (1971) identified the following eight different semantic relations expressed by two-word.utterances in the data he examined: 1. Agent and Action 2. Action and Object 3. Agent and Object 4. Mbdifiers 5. Negation 6. Dative 7. Ostentation 8. Locative He suggested that these intention-markers or I-markers are determined.by'the cognitive capacity of the child, and are universal and innate. Bowerman’s (1973) study of Finnish and English also provided support for a semantic interpretation of the structural relations 4 expressed in children’s early utterances. She noted that in cross- linguistic comparison of children’s speech, there were striking similarities in the constructions of their early word combinations. In 1973, Brown provided a comprehensive review of 19 reports on 13 children Observed by a variety of investigators (Brains, 1963; Brown & Fraser, 1963; Brown & Bellugi, 1964; Miller & Ervin, 1964; Brown & Cazden, 1969; Kernan, 1969; Bloom, 1970; Bowerman, 1970; Rydin, 1971; and Tobert, 1972; cited in Brown, 1973). The data were gathered from the spontaneous conversations of children in natural environments with similar MLU values. The various languages studied included American, English, Finnish, Swedish, Samoan, and Spanish. Brown examined the semantic relations expressed in children’s utterances during his defined MLU Stage 1. On the basis of this review, he was able to identify the following eight basic semantic relations which accounted for approximately seventy percent of the children’s utterances across studies: 1. Agent + Action 2. Action + Object 3. Agent + Object 4. Action + Locative 5. Entity + Locative 6. Passessor + Possession 7. Entity + Attributive 8. Demonstrative + Entity This suggested that a set of basic semantic relations was universal across unrelated languages. Br ob er in 5 Although the notion of the semantic relational category has provided an invaluable base from.which to explore and describe early language developnent, it should not be assuned that these semantic relations adequately represent the complex and multi-faceted nature of language, as the following section will show. This study was an attempt to expand the description of semantic relational categories in children’s language. Statement of the Problem While the semantic relations developed by Bloom (1970), and Brown (1973), were excellent first attempts in the description of semantics of language they have one basic shortcoming. Casual Observation of these semantic relations, reveals that they are extremely "glObal" in nature. The use of global categories results in assigning a wide range of utterances to a single category even though they are very different in meaning. Fer example, the verbs "breaking" and "sleeping" could be classified under the semantic category of action. However, on closer inspection, the meanings expressed by these words intuitively seem to be different. "Breaking" is an action which requires movement but results in an external change of the attributive state of an Object. Hewever, "holding" is an action that neither requires movement nor results in an external change in the state of the person involved. There are other examples of action that code different aspects of meaning, yet a glObal category would classify them as being the same. glc 81] 10c new fro din 5901 and utte diff E § wiric It 3, WC the c 15’le 6 One can also identify various aspects of meaning within the global category of location. For example, "the dog junped off the chair," "the dog junped down," "the dog jumped on the chair," would all be classified under the sanantic category of location or locative action. However, closer inspection of these utterances reveals that each utterance has slightly different shades of meaning from the other. The original location of the dog is highlighted in the first example, "the dog junped off the chair," whereas, the directional location of the dog’ 8 movement is identified in the second example, "the dog junped down." In the third example, "the dog junped on the chair," it is the destination of the dog which is emphasized. Within a global system of semantic categories, these utterances all would be placed in a single category, and the differences in their meanings would not be revealed. These examples point to the possibility that the global semantic categories described by Bloom (1975), Bloom, Lightbown, and Hood (1975), and Brown (1973) can be further differentiated into more fine-grained subcategories which could better describe the various aspects of meaning coded by a global category description. It scans reasonable to assure that more differentiated categories could capture developmental differences. The next section explores the consequences of a global model for 1) the description of later language developnent, and 2) the differentiation of normal and We-impaired children. 7 TheLse of Global Sen_1aintic Categories in the _Description of miter W The basic sanantic relations of Bloom (1970) and Brown (1973), do not adequately describe later semantic development. This conclusion is not surprising given the assunption that semantic relations such as agent, action, location, negation, etc., are assuned to be basic and universal features of language. Therefore, they would be expected to occur in some form in all children at an early age. In fact, these semantic categories were developed as a description of children’s language at early stages of development. The children in the studies reviewed by Brown (1973) were all at a defined MLU Stage I of development which spanned the ages of 1;? and 2;6 years . Brown concluded that Stage I utterances in all languages for which studies exist, concentrate on the same set of meanings, a set far short of the meanings that languages are able to express and in adult usage, do express (p. 173). This would suggest that while Brown's (1973) semntic relations are descriptive of children’s utterances during Stage I of development, they are unable to account for the meanings the child acquires later in development. These results were supported by Stocknan and Vaughn-Cooke (1982) in their canparison of data on working class Black children, with the data collected by Bloom, Lightbown, and Hood (1975), Miller (1982), and Blake (1984) on working class and middle class White children. Their review indicated that the mjority of the children 8 in Mid Stage I (19-23 months) had reached the criterion of productive usage for the following semantic categories: 1. Existence 2. Action 3. Locative Action 4. State 5 . locative State 6. Negation 7. Passession 8. Attribution 9. Notice 10. Intention 11. Recurrence Because these basic semantic relations are acquired early in language development, one is left with the question of what happens later in the development of language. Do children simply add more forms once a category is acquired, or is there a qualitative difference in the type of forms acquired at various stages of development? If glObal categories of semantic relations cannot distinguish among normally developing children beyond the earliest stages of language development, then one might wonder about their effectiveness in identifying semantic differences between language- impaired and normally developing children. 0f def e". i 1 9 The Use of Global Segntic Categories in the Differentiation of Nornal and We wired Childreg Language assessment in clinical contexts base the determination of norual versus language—impaired on nornal language developmental data. The assunption that global senantic categories are unable to adequately describe differences among normally developing children predicts that they also will not differentiate between norual and delayed development. The absence of a differentiated way of looking at semantic relations, therefore, has implications for understanding the nature of language-impairment. The cognitive/representational hypothesis is among current views regarding the nature of language-impairment. The fundamental premise of this hypothesis is that a cognitive deficit is at the base of language-impairment. Language has been thought to be related to cognition prinarily in the semantic system within contanporary interpretations of this relationship (Vygotsky, 1962; Bloom, 1970; Slobin, 1973; Schlesinger, 1974). This line of argunent suggests that if cognitive deficits exist among the language-impaired, then the senantic system should reflect these deficits. While there has been a great deal of support for the presence of cognitive deficits among language-impaired children, the literature has not provided support for a corresponding semantic deficit. The remainder of this section will first consider the evidence suggesting that cognitive deficits nay be characteristic of 0h 10 language-impairment, and then the evidence suggesting that semantic deficits may not be characteristic of language-impairment. Evidence Suggesting Cognitive Deficits May Be Characteristic of ngguage Impgirment Several researchers have attempted to investigate the cognitive abilities of language-disordered children by comparing their performances on various nonverbal tasks to the performances of age- natched children with normal language development. In spite of documented normal nonverbal intelligence, the language-impaired groups performed more poorly than the normal groups in the areas of anticipatory imagery (Savich, 1984), mental rotation abilities (JOhnson & Weismer, 1983), symbolic play (wain & Yule, 1983), haptic recognition (Kamhi, 1981; 1984), discrimination-learning prOblems (Nelson, Kamhi, & Apel, 1987), and cognitive and semantic processing (Wren, 1982). These results suggest that nonverbal cognitive deficits do exist in children with specific language- impairments. This notion of related nonverbal cognitive deficits in children with language disorders is at variance with current definitions of specific language impairment. Specific language impairment is defined as a delay in a child’s comprehension and production of language in the absence of any sensory, perceptual, emotional, or cognitive impairment (Stark & Tallal, 1981). Language-impaired children are different from mentally-retarded children, in that they 11 don’t have a general cognitive deficit, and it isn’t known why the language problem exists. This discrepancy between research findings that support a cognitive deficit in language-impaired children, and the perception that they have nornal cognitive abilities nay be related to the way that cognition has been measured. Nonverbal cognitive abilities are often treasured by intelligence scales such as the Egg; Intem_ation_a_l Perforgaince Sca__l_g (LIPS, Leiter) (Leiter, 1959). Johnston (1982) analyzed the LPS items and found that they fell into two groups depending on the types of cognitive processes that they entail. One group included perceptual items which required only the recognition of physical resemblance, and the other group included conceptual items, which required that the picture be interpreted according to prior spatial, ntmerical, or classificatory knowledge. Johnson administered the LES to 16 language-disordered children matched to 16 children with normal language, on the basis of chronological age, sex, and Leiter IQ. The results revealed that the language-disordered group was more successful on those itans requiring only the perception of physical similarity. Therefore, language-impaired children nay exhibit deficits in other areas of nonverbal cognitive functioning although their visual perceptual processing skills nay be age appropriate. If this were the case, these children would receive age appropriate scores on tests such as the LIE because the test did not tap into the nonverbal cognitive functions with which the child has difficulty. 12 Thus, the notion that specifically language-impaired children have normal nonverbal cognitive abilities is questionable, given the possibility that inadequate measures of cognition have been used. The hypothesis that cognitive deficits are at the base of language- impairment has received support in the literature. If cognitive deficits are at the base of language-impairment, then semantic deficits should be a characteristic of language-impairment. However, studies do not support this expected corresponding deficit in the semantic system of language-impaired children, as the following section reveals. Evidence Suggesting Semagtic Deficits May Not Be Characteristic of W A review of the literature reveals that there are relatively few studies comparing the semantic relations used by normal and language-impaired children. Although these studies were few in number, all convincingly argued that semantic deficits were not evident in language-impaired children. One such study was conducted by Freedman and Carpenter (1976), in which the semantic relations expressed by four language-impaired children were compared to those of four nonmally developing children matched on the basis of level of linguistic development, social position, and sex. All children were Stage 1 according to Brown (1973) as determined by an MLU of 1.4 to 2.1 morphemes. Three hundred non-imitated two-word utterances were collected from each child and categorized into one of the following semantic relations: 13 introducer + entity, more + entity, negation + entity, agent + action, action + object, action + locative, entity 4- locative, possessor + possession, and entity + attribute. A significant difference was revealed in only one relation, introducer + entity, with the language-impaired group demonstrating greater diversity in usage than the nornal group. NO other significant differences were obtained between the two groups on the renaining nine relations. These results indicated that, at the Stage I level of language development, the language-impaired children demonstrated at least as much flexibility in their usage of different semantic relations as their language-matched peers . Fokes and Konefal (1981), examined the use of the case relations, "agentive," "action," "objective," and "locative" in seven language-disordered children (ranging in age from 5;0 to 7;0 years) compared to ten nornal age-matched children (ranging in age from 5;0 to 6;0 years) and six nornal younger children (ranging in age from 3:0 to 4;O years). The case relations were elicited by having the children describe both observed activities and self- nanipulated activities to a blirndfolded doll. Their results indicated that the language-disordered group produced more single and two-word utterances than the three and four case relations. However, all four case relations were produced by the language- impaimd nonp- In 1976, Leonard, Bolders, and Miller compared the seuantic relations reflected in the language usage of 10 langmge—disordered and 10 nornal children ranging in age from 2;11 to 4;2 years, and 10 14 language-disordered and 10 nornal children ranging in age from 4;8 to 5;8 years. A 50 utterance language sample was obtained from each subject by asking five standard questions to elicit stories about 10 different pictures. The utterances were categorized using the semantic relations: agentive, instrunental, dative, locative, objective, and essive. The results revealed that when the subjects were natched for age, the nornal and language-impaired subjects did not differ in the type of semantic relations used by each group, but some differences in the frequencies with which semantic categories were produced, were evident between groups . When the subjects were natched for mean length of utterance, the nornal and language- disondered subjects did not differ in the frequency with which their language usage reflected the different senantic relations. Duchan and Erickson (1976) investigated the semantic relations produced in twelve normal and twelve language-disordered mentally- retarded children natched for mean length of utterance. A 60—item comprehension test was developed in which the subjects were required to nanipulate various familiar objects in response to verbal stimuli. The senantic relations agent-action, action-object; possessive, and locative, were equally represented. No significant differences were found between the mentally-retarded children, and the normally developing children in their comprehension of these four seuantic relations. Coggins (1979) examined the senantic relations produced by four Down’ a Syndrome mentally-retarded children, two of whom were placed into Early Stage 1 on the basis of MLU’s of 1.25 - 1.50 morphemes 15 and two of whom were placed into late Stage 1 based on Mills of 1.60 - 2 .00 morphemes. Two-word non-imitated utterances were transcribed from language samples and classified using the following semantic categories: demonstrative—entity, negation-entity, agent-action , action-object, action-locative, entity-locative, possessor- possession, and entity-attribute. The results revealed that all subjects encoded at least a few instances of each relation while most of the categories were represented by a large number of different two-word utterances. Thus, it appears that Down’ a Syndrome children at Stage 1 of linguistic development concentrate on the same small set of relational meanings as in normal children’s early two-word combinations . In summary, these studies indicated that language-impaired children encode the same semantic relations as do their MLU matched peers. This has been demonstrated in both specifically language- impaired children and mentally-retarded children. In fact, Leonard, Bolders, and Miller (1976) did not even find significant differences in the type of semantic relations used by age-matched language- impaired and normally developing children. It is interesting to note, however, that all of these studies examined semantic categories using global frameworks. Possibly a more differentiated set of subcategories would be able to reveal semantic deficits in language-impaired children. Until more defined subcategories are developed and applied to language-impaired children, the question of whether semantic deficits are characteristic of language-impairment will remain unanswered. fa t: 16 This study attempted to shed some light on this issue by expanding the description of one global semantic category -- namely the category action. The following section will discuss the definition of action as viewed by various researchers and why it is important to look more closely at action. Definition of Action Many definitions of "action" include the notion of volitional or purposive behavior (see for example, Brandtstadter, 1984; Huttenlocher, Smiley, & Charney, 1983; Chafe, 1970). In his conceptualization of verbs as states, processes, and actions, Chafe discussed action as "something which someone does" (p. 100). Within this perspective then, an action utterance requires an agent, and is able to answer the question, "What did the agent do?" However, classifying verbs according to the type of noun they take (i.e. agent, patient, instrument, experiences, etc.) seems to be a reflection of the semantic meanings of nouns, rather than a reflection of the meaning of the verb itself. To classify "falling" as an action in the sentence, "the man is falling," and as a process in the sentence, "the blocks are falling" because "man" is an animate Object, and "blocks" are inanimate patients, disregards the fact that "falling" in.both instances is recognizable as the same type of movement, with the same resulting change in location. In both instances, the Objects are capable of independent movement. Movement and change are not always under the control of animate agents. The forces of nature can also cause things to move or he uh, do: lit, 17 change as in the examples, "the leaves are falling," "the child is growing." These statements refer to happenings or events, and can answer the question, "What is X doing?" where "X" is a noun. Bloom arnd Lahey (1978) did not include the notion of intentional behavior in their definition of action. Action referred to voluntary or involuntary movement that affects only the person or object engaged in the movement or both the object engaged in movement and another person or object. The concept of including only those verbs involving movement in the category of action is not consistent with Chafe’s (1970) discussion of action as "something which someone does." Verbs such as "sleeping, sitting, or thinking" obviously are things which someone can do, but do not require movement. While specific points in the definition of action are not consistent across various interpretations, there appears to be general agreement that action refers to happenings or events as opposed to conditions or states of being. Justification for Lookigg at Subcategories of Action It seems appropriate to select action as the semantic relational category of choice. Gnildren talk, overwhelmingly, about what they are just about to do, what they are doing, or what they are trying to do and, less often, about what they see other people doing (Bloom & Lahey, 1978). This notion has been supported in the literature by Rodgon, Jankowski, and Alenskas (1976), and 18 Huttenlocher, Smiley, and Chamey ( 1983) . Several studies have indicated that the earliest words to develop in children refer to moving objects (Huttenlocher, 1974; Nelson, 1973). Thus, it appears that the child’s performance of action is a crucial factor in the child’s development of receptive and expressive language skills. In addition, action has been prominent in theories of cognitive development, and in discussions of language development in both normal and clinical children. The following sections call attention to the role of action in these areas. Action asfiRepresented in Cognitive Development Action is central to Piaget’s theory of intelligence. Piaget (1963, 1970, as cited in Morehead & Morehead, 1974) maintains the primacy of action over perceptual and symbolic structures as the primary contact with, and organizer of reality. According to Piaget’s theory of sensori—motor development, the child under 1;6 years of age learns by applying action schemes to experiences in his world. In the process of manipulating things, they are transformed from an existing state to an alternative state, and as a result of the transformations, the child comes to discover the properties and relations of objects and events in reality (Morehead & Morehead, 1974). Perception is regarded as necessary only for recognition, whereas actions are necessary for understanding or meaning. The result of acting on objects and with objects is that children become aware of their relationships to one another, and in relation to their own actions. In this way, the child is able to organize his 13> 19 mental schemes arounnd relations among objects. Therefore, it is important that action, in the form of sensorimotor schemes is the principal mode by which the child interacts with his world. In Nelson’s (1982, 1986) revised Emotional-Core Model of natural concept formation, the importance of action in conceptual and language development also is emphasized. She suggests that children’s conceptual development is built around "event structures" which are basically sequences of actions. The "event structure" and not the "object structure" is primary, because objects are first knnown in their relation to the events of which they are a part. Nelson argues that, while perceptual information is important to language learning, little language learning would occur if the child were presented with just the pattern of objects. "The place of the object in the pattern of activity represented by the child’s event scripts, needs to be established for the child to confer meaning on the object within his or her own conceptual system" (p. 353). Thus, the basic form of conceptual representation is that of event representation involving sequences of action. In both Piaget’s (1963, 1970, as cited in Morehead 8:. Morehead, 1974) and Nelson’s (1982, 1986) theories of cognitive development, action plays a critical role. Action as Represented in Normal Me Develognt Because of the important role that action plays in the child’s early cognitive development, it would be natural to assnme that words coding action would dominate early language development. Inc ”C Ch 20 While it is generally agreed that the action is one of the first semantic categories to develop in young children (for example, Nelson, 1973; Benedict, 1979; Fokes & KOnefal, 1981), a considerable body of research has reported that children learn a large number of object names when developing language (Nelson, 197 3; Goldin-Meadow, Seligman, & Gelman, 1976; Leonard, 1976; Huttenlocher & Lui, 1979; Schwartz & Leonard, 1984). These observations suggest that Objects, not actions, are important to the child’s concept of the world. If action, as proposed, is at the base of development, one might wonder why action words do not dominate early vocabularies, as do Object words. In addressing this issue, Nelson (1982) offered three explanations. First, the nature of communication interaction between parent and child lends itself to the labelling of objects. Adults may teach children to attend to Object names because Objects are more concrete to point out to children, than are actions. The attention of a parent and/or child can focus more easily on Objects than on actions. In.addition, it appears that there are many more nouns than verbs in the adult language system and therefore one would expect nouns to appear more frequently than verbs in the child’s first vocabulary. Nelson (1982) argues, the fact that object terms are learned predominantly does not reflect the fact that Objects are a predominant part of the conceptual representation, but rather that they are a salient aspect for the adult and child to focus upon in an on-going action sequence (p. 356). Nelson’s second explanation for the predominance of Object words in children’s early vocabulary, is that the child may be using 11 If? St. in dev. 21 an object label to refer to any part of a situation, not simply the object itself. In early language development, many children appear to use a word to refer to any aspect of a given situation, or to a given function rather than to the object. Thus, while adults may be labelinng the object within a given situation, the child may connceptualize that label as representing the action being carried out by the object. A third explanation for the predominance of objects in early lexical acquisition, as put forth by Nelson (1982), is the fact that objects may be more variable than actions within a given situation. As the child attempts to decontexttalize aspects of a situation, "objects become named, because objects are variables. Actions, however, are specified by the situational context" (p. 357). Thus, while action words may not dominate early developing lexicons, the concept of action appears to be critical in the child’s cognitive and language development. Action a; Represented in [me—Mgirment The close relationship between the child’s ability to act on the environment and his early conceptual development has provided a theoretical base for interesting research currently taking place in St. Gallen, Switzerland. Dr. Felicia Affolter at the Center for Perceptual Disturbances in St. Gallen, Switzerland and a multidisciplinary team have developed a treatment framework for the language-impaired within 13 22 which perception plays a critical role (Stockman, 1986). They propose that not only must the environment present new situations to the child in the sense that they offer prOblems to be solved, but the child.must have enough sensory information to perceive the situation as having an unfamiliar aspect, and therefore presenting a prOblem to be solved. From all sensory input associated with successful prOblemrsolving activity comes knowledge about the worlds-the functional properties of Objects and their relationships, how to plan events, change and reconstruct them, and finally, what aspects of events are encoded through language. Therefore, prOblem-solving exploratory activity is viewed as the developmental root for verbal and nonverbal behavior (p. 16). The St. Gallen team argues that developmental problems for nanny children can be traced to their inability to explore the environment adequately for learning, because of perceptual handicaps. Further, their research has shown that tactile-kinesthetic deprivation and its lack of central integration with other sensory systems has a more adverse impact on the learning of complex skills than do visual or auditory deprivation (Stockman, 1986). The strong link between action and cognitive development (Piaget, 1952), and action and language development (Rodgen et al., 1976; Bloom & Lahey, 1978; Huttenlocher et al., 1983), support St. Gallen’s theoretical view of learning. §HEE§EX The shift toward the semantic description of language provided a better description and a greater understanding of children’s language acquisition in the earlier stages. unfortunately, the semantic categories of Bloom (1970), and Brown (1973) are too b: 23 "global" in nature to identify developmental patterns within categories or to describe later stages of language acquisition. It is possible that if more refined subcategories of semantic relations were developed, more could be learnned about normal acquisition of semantic categories, and consequently provide a better understanding of language-impairment. It would seem particularly valuable to devise a system of subcategories of action because of the central role that action plays in some theories of cognitive development and in discussions of language development and language-impairment. In fact, the semantic category of action is one of the earliest categories to emerge, and although action words do not dominate early lexicons, there are very good reasons for this, as provided by Nelson (1982). If a system of action subcategories were devised, it could be used in developmental studies of language acquisition to identify trends in the acquisition of action relations. Given the centrality of action in the organization of a child’s symbolic system, action would likely be a target for assessment in clinical settings. Thus, a system of subcategories of action could also provide pertinent information that would be valuable for a language assessment. HOwever, the first step is to develop a system of subcategories of action which can describe the majority of action utterances produced by children. 24 Purpose Therefore, the purpose of this study is two-fold: 1. Can a set of action subcategories be developed from the Observation of the action relations expressed by normally developing children? 2. Can the application of such action subcategories reveal differences between normal and language—impaired children? in de 21 it SU thi REVIEW OF THE LITERATURE This study has been based on the premise that the semantic categories devised by Bloom (1970) and Brown (1973) are too global in nature to provide an adequate description of normal language development, or language impairment. This study focused on the glObal semantic relational category of action to determine whether it can be described in terms of a set of more differentiated subcategories. It is therefore relevant to take a look at other attempts to identify semantic subcategories to determine whether such a notion has yielded useful outcomes. This chapter reviews those studies that have described subcategories of semantic relations. The notion of subcategories of semantic relations is not a new one. various subcategories of semantic relations have been developed both from a linguistic perspective on the basis of the adult language system and within the context of normal language acquisition. Subcategories Based on the Adultngggggg§_§y§§gg Because this study focused on the subcategorization of action, this section considers those works dealing exclusively with this semantic category. Chafe (1970), Edwards (1974), and Muller and 25 John cats 100) 838) DORE sem typg the] The 26 Johnson-Laird (1975) have attempted to subcategorize the semantic category of action, based upon the adult language system. Chafe (1970) reflected upon the semantic nature of verbs by looking at the semantic nature of the nouns accompanying the verbs. He noted that verbs specifying state are accompanied by a noun which is a patient. For example, in the sentence, "the wood is dry", the patient (wood) is said.to be in a certain state or condition (dry). Chafe distinguished nonstate from states by the fact that nonstates answer the question "What happened?", "What’s happening?", etc. A nonstate was defined as a "happening", or event. For example, the sentence, "the wood dried", can answer the question "What happened?", whereas the sentence "the wood was dry", cannot. Chafe (1970) went on to point out major differences between two types of nonstate sentences. He identified verbs as processes, wherein the noun is said to have changed its state or condition. The nouns in these sentences are also the patient of the verb. For example, the sentence "the wood dried" describes a patient (wood) which has undergone a process (dried) resulting in a change of state. Chafe (1970) also identified verbs which expressed an activity or action, something which someone does, that has nothing to do with something which performs the action, The nouns in these sentences are agents because they specify either a state or a change of state. For example, "the man laughed" describes an agent (man) who performed an action (laughed). Chafe distinguished the actions from processes in that an action sentence will answer the question "What did N do?", where N is somennoun and often a simple process sentence 27 will answer the question, "What happened to N?" For example, the question "What did Harriet do?" can be answered appropriately by the sentence "she sang" (action) but not by "she died" (process). Conversely, the question "What happened to Harriet?" can be answered appropriately by the sentence, "she died." (process), but not by "she sang" (action). Chafe (1970) further described sentences in which the verb is, simultaneously, both a process and an action. In these instances, the action is expressed by what someone, its agent, does, and the process involves a change in the condition of a noun, its patient. For example in the sentence, "Harriet broke the dish" the agent (Harriet) performed an action (broke) resulting in the change of state (broke) of the patient (dish). Edwards (1974) discussed a systematic way of organizing verbs in terms of two orthogonal dimensions, which were: 1) the type of state or relation (locative, possessive, attributive); and 2) whether the given state or relation was an unchanging state of affairs, one that involves change or one that is caused to change or happen by an agent and/or instrument. verbs were classified as either "actional ver " which lack any inherent specification of any necessary change of state or of spatial position of the Object affected, and "stative verbs" which describe a state of affairs or a changing state of affairs where the establishment of a new state is implied. The "actional verbs" were further subcategorized into "direction action verbs" such as "hit," "stroke," "punch," which describe a type of physical contact between instrument and object, 28 and "movement verbs," such as "roll," "turn," "walk," which describe the type of motion or activity gone through by the Object. "Static ver " were classified as "possessive" (e.g., buy, own, give), "locative" (e.g., on, enter, evacuate), and "attributive" (e.g., fat, break, shattered) depending upon the type of state of affairs they describe, or in which they describe a change. Edwards applied this classification system across static, dynamic, and causative events . The dynamic, and causative events were distinguished by the fact that the nouns in dynamic events are instruments, Objects, or experiences, whereas the nouns in causative events are always agents. In 1975, Miller and Johnson-Laird.attempted to subcategorize action in their discussion of verbs of motion, possession, vision, and communication. verbs of motion (e.g., walking) were described as "how people and things changed their places and their orientations in space" (p. 527). verbs of possession (e.g., give) were primarily a conceptual matter going beyond perception (p. 588), in contrast to verbs of vision which referred to the sensory modality of perception (p. 601). verbs of communication (e.g., saying) were considered verbs that "talk about talking" (p. 619). Unfortunately, Chafe (1970), Edwards (1974), and.Miller and Johnson-Laird (1975) devised systems of classification which were theoretically based without empirical data to determine their effectiveness in revealing developmental change in children’s language . 29 subcategories Based on Normal Laggugge Development subcategories of semantic relations developed within the context of normal language development have included the differentiation of the glObal semantic category negation, location, and action.1 These subcategories have proven effective in revealing the developmental changes among children. Subcategories of Negation In 1970, Bloom redefined the category of negation into the subcategories of nonexistence, rejection, and denial. Nonexistence was coded when "the referent was not manifest in the context, where there was an expectation of its existence" (p. 173). Rejection was coded when "the referent existed or was imminent within the contextual space of the speech event and was rejected or opposed by the child" (p. 173). In instances where "the negative utterance asserted that an actual (or supposed) predication was not the case," denial was coded (p. 173). Far example, the utterance "no doggie" could express rejection if the child wished to play with the toy cat and was given the toy dog, or denial if the child was given the toy cat and told that it was a toy dog. Nonexistence would.be coded in the utterance "no 1Bloom and Lahey (1978) also discussed.categories of state relations as including possessive state (e.g., the book is mine), attributive state (e.g., the hat is brown), internal state (e.g., she likes ice cream), and external state (e.g., it is dark). However, developmental study of children’s acquisition of these categories was apparently not completed. Therefore, they will not be elaborated further. 30 doggie" if the dog a child was playing with walked out of the room. While these utterances can all be coded within the global semantic category of negation, it is clear that the subcategories are able to differentiate between the more subtle aspects of meaning coded in these utterances. Within the course of a child’s language development, Bloom (1970) found that the order of acquisition for negation was specified as nonexistence, rejection, and denial. subcategories of Locggion More recently, Stockman and vaughn-Cooke (1984, 1987) and Stockman (forthcoming) have investigated subcategories of locative utterances. The following eight locative subcategories were identified: 1. static origin 2. static direction 3. static destination 4. static combinative 5. dynamic origin 6. dynamic direction 7. dynamic destination 8. dynamic combinative. A locative utterance could be dynamic in which movement is coded, or static in which no movement is taking place. A locative expression (dynamic or static) could exist in terms of a place of origin (dynamic: the ball fell off the table, static: shells come from the ocean), the direction of movement (dynamic: the cat jumps 31 down, static: the college is south of here), a place of destination (dynamic: I set it on the table, static: it is on the table), or any combination of the above three terms (dynamic: I set it down on the table, static: it’s down on the table). When these subcategories of locative utterances were applied to children’s normal development of language, the developmental nature of the subcategories was revealed. In the category of locative action children talked about the source from which Objects move or the path of movement before they talked about the destination to which Objects move. In addition, they used one locative word to refer to one aspect of the locative event before they combined locative words to refer to more than one locative aspect of the same event. Within the category of locative state, children talked about the immediate position of an object before they talked about the object’s position from the directional perspective or from the perspective of a former locative site. Subcategories of Action Gentner (1978) and Huttenlocher, Smiley, and Charney (1983) attempted to identify subcategories of action in children’s language. Gentner (1978) addressed the issue of action subcategories by examining the "subpredicates" of verbs which are intended to express the almost inevitable inferences made in verb comprehension (p. 989). In particular, she investigated the action verbs coding change in possession: "give," "take," "pay," "trade," "buy," "sell." She asked children aged 3;6 to 8;6 years to act out sentences such as "Make Ernie buy a car from Bert," using dolls with 32 toys and money. Her results indicated that the verbs were acquired in order of complexity, with the simpler verbs "give" and "take" being acquired earlier than the more complex verbs "buy" and "sell." For example, all the meaning components of "give" are also contained within the representation of "sell." While this study recognized that subcategories of action exist, Gentner chose to investigate the acquisition of one subcategory rather than identify a variety of subcategories. Huttenlocher, Smiley and Charney (1983) attempted to identify broader classes of subcategories of action. They categorized the verbs that young children produce into verbs coding change and verbs that did not code change. "Nonchange" verbs were defined in terms of characteristic movements by an initiator (e.g. walk, wave), whereas "change" verbs (e.g. open, get) were defined in terms of changes caused by an initiator. The context was also considered in terms of "self action," defined as when a person acts, as a subject experiencing a goal, and "observed-action," defined as a person as an observer witnessing movement by an initiator. They conducted their study in three parts. Part 1 of the Huttenlocher et al. (1983) study examined the comprehension of verbs in relation to observed action. They randomly presented 69 children ranging in age from 1;10 to 3;6 years, ten pairs of verbs, four times within the test. Each action was presented within a videotape stimulus. The experimenter named both actions and then asked the child to identify the target action. ir. ch it 0‘. 33 Results indicated that children apply movement verbs to observed behavior earlier than they apply change verbs. In Part 2 of their study, Huttenlocher et al. (1983) examined the contexts in which sixteen children ranging in age from 24 to 28 months, used verbs. Four hours of spontaneous production data was Obtained from each child during a normal day’s activities. Of the 1,066 utterances with verbs, 90% were produced when the child was participating in the action in some way; The children rarely used verbs to encode observed behavior. Another interesting finding was that, while the subjects did not describe actions of others that involve change, they did describe their own actions involving change. On the contrary, while the comprehension data indicated that children use movement verbs to encode observed action, children did not spontaneously produce these verbs to describe either their own movements or Observed movements. The purpose of Part 3 of the Huttenlocher et al. (1983) study was to determine whether verbs that encode characteristic motions, like verbs that encode change, apply first to the child’s own actions. Ten children were followed longitudinally, beginning when they were one year old and ending when their MLU’s were 2.3. During monthly visits, the children were tested in their comprehension of the verbs "sit down", "run", "kick", "jump", and "wave bye bye" in relation to self and others. In addition, 14 of the children were also tested on the verbs "put down" and "get" in relation to self and.others. Each verb was presented as an instruction and in the movie task used in Part 1. The children were tested.each.month 34 until they succeeded on one of the tasks, three times in succession. The results indicated that verbs encoding movenment are also acquired first for the child’s own actions. In summary, results of the Huttenlocher et al. (1983) study, indicate that children comprehend and produce both verbs encoding movement and verbs encoding change in relation to their own action before Observed action. verbs of movement are comprehended earlier than verbs of change, in both contexts of self-action and Observed— action. However, verbs encoding movement are rarely produced in either context. While these authors have provided some basis for looking at subcategories of action, casual inspection of the categories proposed by Gentner (1978) and Huttenlocher, Smiley, and Charney (1983) suggests that they will not account for all the action utterances produced by young children. For example, "standing" cannot be categorized within verbs of possession or change. While it appears that "standing" could be classified as a nonchange verb under the Huttenlocher et al. (1983) system, it does not meet their definition of coddng a characteristic movement. It appears that a .more comprehensive system of subcategories has yet to be developed which can account for all the meanings coded in children’s action relations. M The investigations utilizing subcategories of semantic relations have provided a much more detailed and accurate 35 description of sennntics than was accounted for within more global systems. Developnental trends, that had previomly gone undiscovered, were revealed by the use of these subcategories in the description of normal language acquisition. These results would suggest that the differentiation of global semantic relations into more refined subcategories is a promising area for further investigation . While the studies examining subcategories of negation (Bloom, 1970) and subcategories of location (Stockman, 1986) encompassed all aspects of their respective categories, those studies examining subcategories of action (Gentner, 1978; Huttenlocher, Smiley, & Charney, 1983) analyzed only restricted aspects of action. A more comprehensive system of subcategories of action has yet to be developed . The fellowing section outlines the procedures employed to answer the questions posed in this study. The first part of this section discusses the procedures used in the development of a set of verb subcategories that could account for the majority of action utterances produced by normally developing children. The procedures employed in the application of these verb subcategories to a comparative analysis of the action utterances of language-impaired and normally developing children, are outlined in the second part of this section. Part I - Action verb Subcategories in a Normative Sample A set of verb subcategories was developed to describe the range of the action relations expressed.by normally developing children between 4:3 and 4:6 years. A data base containing cross-sectional/ longitudinal data on children’s spontaneous utterances was utilized in this endeavor. The data were collected in Washington, D.C., between December, 1980 and June, 1982, by Stockman and vaughn—Cooke (1982) for the purpose of studying a wide range of developmental linguistic issues that require naturalistic data sampling. Naturalistic spontaneous language samples were desirable for this study for two reasons. First, in order to obtain as valid a 36 37 picture as possible, it was necessary to look at children’s language when no restrictions were placed on what they could talk about. Until we learn about children’s naturalistic language, we will not know what types of information would be worthwhile to highlight in a .more restrictive, experimental fashion. The value of spontaneous data for the initial investigation of semantic relations had been demonstrated by the classic work of Bloom (1970), Brown (1973), and Bowerman (1973), as well as many others. Second, spontaneous language samples seem to be an effective method of collecting data specifically on action verbs. Research has indicated that children primarily communicate about their own actions (Rodgen et al., 1976) and use action verbs to code their own actions before they use them to code the actions of others (Huttenlocher et al., 1983). Description of the Data Base General subject Characteristics of the Data Base The twelve subjects in the data base were children from working class families in the washington, D.C. area. The longitudinal database extended over an 18-month period and represented three cross-sectional age groups. Three groups of four children, with two boys and two girls in each group, were the ages 1:6, 3:0, and 4:6 years at the beginning of the sampling. At the end of the sampling period these groups of children.were 3:0, 4:6, and 6:0 years, respectively. The children were selected from families affiliated with Headstart Programs in the Washington, D.C. areas The school personnel were required to provide information regarding children’s 38 health status, history, educational progress, and general functioning, using standard written questionnaires. The subjects used in the data base were then randomly selected from among the children who met the normalcy criteria. Procedures Used to Collect the Data Base (he- to two-hour language samples, containing at least 400 utterances, were collected at four- to six-week intervals from each of the twelve children over an eighteen—month period. Audio-visual records were made of each language sample . In order to represent as clearly as possible the equipnent used to collect the data base, a direct quotation has been taken from Stockman and Vaughn-Cooke (1984) . The field equipnent consisted of a portable color camera (JVC—G-7IUS) equipped with view finder and automatic light control and a video-cassette recorder (Sony SID-323) . Video clarity was neximized by supplementing home lighting with high voltage lamps where required. Further, a portable television monitor provided continuous onsite feedback about video quality. A tie tack microphone (EGG- 31 with frequency response range of 50 to 13,000 Hz) was attached to the child’s clothing - typically the collar - at distances permitting clear and undistorted audio quality. The microphone and camera cables permitted the child to move freely within an eighteen foot area. Samples were obtained during routine play activity involving the subject in social interaction with children and adults, including an investigator. A core set of toys including a doll house, basic house furniture, assorted wooden blocks, a ball, balloons, etc. , was used with all the children, in an attempt to facilitate comparability of data among the children with respect to what they talked about. The child was encouraged to play with toys 39 and Objects in his own environment, in addition to the toys provided by the investigators. Little or no structure was imposed on the child’s actions during the sampling. They were free to play with, and talk about, whatever they wished (p. 12). In all, approximately 75,000 utterances for analysis were included in the data base, spanning a relatively early to late developmental period of 1:6 to 6:0 years of age. This provided a substantial corpus from which action utterances could be selected for analysis. Selection of Data for Study of Action Subcategories For this study, the utterances extracted from the data base included the first hour of the language sample taken from eight children (4 boys, 4 girls) at the age of 4:3 to 4:6 years. For four of the children, the data represented the first sampling period, and for the remaining four the data represented the eighteenth sampling period. As estimated, this selection of data provided 1,627 utterances (or approximately 200 per child) from which to select action utterances for analysis. Treatment of the Data for Analzais Form of the Dataafor Analzais The data were available in two forms. Auiio-visual records of the language samples were available in color on Beta video cassettes. The audio-visual record provided the situational and linguistic context in which the child’s action utterances were 40 produced. It was determined.what was said before and after each of the child’s utterances and what the child and other participants were doing before, during, and after the utterances. This was very helpful in interpreting the meaning of the child’s utterances. For example, the utterance "the boy go here" produced within the context of the child moving a toy doll from one position to another could code action. On the other hand, the same utterance, "the boy go here," produced within the context of the child pointing to the toy doll could code state when interpreted as "the boy belongs here." The language samples were also available in written form. All utterances had been transcribed orthographically onto standard forms. The standard form provided space for the utterance number, counter number, contextual notes, the utterance itself, and the semantic categories represented by the utterances. General and specific contextual notes for locative action utterances were transcribed (see Appendix A for a sample of the transcript form). Procedures for Identifyiqg Action Utterances The utterances coding action were identified by reviewing the audio-video tape in conjunction with the written form. The written formtwas helpful because it provided one investigator’s interpretation of what words the child was saying. .A prior utterance gloss was particularly useful if the child was not very intelligible. A review of the audio-visual tape helped to confirm the transcription of the language sample and ensured that no action utterances were overlooked. Equally important, the audio-visual 41 record provided the communicative and situational contexts needed to verify action meaning of the utterances. Qpegaiional Definition. It is generally assuned that action information is explicitly carried in the verb of a sentence (see for example, Bloom, 1970: Chafe, 1970). For example, in the sentence, "the boy runs to the store ," the action performed by the agent is coded in the verb "runs." In this study, an utterance was considered to code action if the following criteria were met: 1. the verb referred to a happening or performance; 2. the utterance was able to answer the question, "What did X do?" or "What is X doing?", where "X" is an agent or Object. This operational definition focussed on the act itself rather than the cause of the act. It reflected elements of both Chafe’s (1970) and Bloom and Lahey’s (1978) concepts of action. Chafe (1970) suggested that nonstate verbs referred to a happening or performance. He then differentiated.nonstate verbs into actions and processes. Action was defined as "something which someone does" (p. 100), and therefore required an animate agent. The notion that an action requires an animate agent was not included in the operational definition applied in this study, because an action event has similar recognizable features regardless of whether it is caused by an animate agent (cf., "man falls" vs. "rock falls"). Therefore, the operational definition of action used in this study included the happenings or performances of both agents and 42 objects. This is consistent with Bloom and Lahey’s (1978) definition of action which did not restrict action events to those having animate agents. Action could be voluntary or involuntary. While voluntary and involuntary action was incorporated into this operational definition, Bloom and Lahey’s (1978) concept of movement being necessary for action was not. The rationale for this decision was that many verbs not involving movement still refer to happenings that are different from state events. For example, "to stand", "to halt", "to hold", do not necessarily involve movement, but require volitional control over their performance. See further discussion on pp. 16-17. Procedures for AssigniagiSubcategories of Action Utterances In applying the operational definition to the identification of action verbs, the situational context of each utterance was considered to determine appropriate meaning. Specifically noted was who or what the child was referring to, if a movement occurred, and if any change of state occurred, or was possible to occur as a result of the child’s actions. The literature (Chafe, 1970: Edwards, 1974: Muller & Johnson- Laird, 1975: Gentner, 1978; Huttenlocher, Smiley, & Charney, 1983) suggested that the framework sunnarized in Table I offered a useful starting point for the subcategorization of action utterances. This framework consisting of 8 subcategories of action verbs, reflected movement and change features of action. Each of these features is described below. 43 Movement/Nonmovement Features Action verbs Codiag Movement. The action utterances in this category involved an overt or observable movement event. That is, some type of movement was necessary in order for the action to occur. The utterances included verbs that coded varying degrees of movement which ranged from movement of the entire body, as in "I am V runniag home,‘ movement of an extremity, as in "She is cuttiag the cake," to small movements of the head or facial features, as in "look at the dog." Action Verbs Codingayogpoveaent. The action utterances in this category did not involve any overt or Observable movement events. That is, the action was able to occur without any movement at all. "She is sleepiag" would be an example of a nonmovement verb because no Observable movement is involved in the act of "sleeping." Although some movement may occur throughout the act of "sleeping," this movement is not necessary to, and does not characterize, the act of "sleeping" itself. Within these two broad categories, action utterances were differentiated further in terms of whether or not the action resulted in a change of Object state. gaaage of State/Nonchaage of State Featurea Action Verba Codigg a Chaage of State. The action utterances within this category involved actions that caused a change of an object’s state. This change may have been in the Object’s physical or attributive appearance (e.g., "Close the door"), existence (e.g., 1:2 act pro of sub. "He "hit. char the Obje COde exte exam eXte: "He j Utter n Sub‘) Wed 44 "§£Qp that"), internal state (e.g.,"I learned that at school"), location (e.g., "I am.driviag the car"), or possession (e.g., "giga me the ball"). In those instances in which the verb did not take an object (e.g., "He is walking"), the agent was affected as a result of the action. These cases could be identified because the reflexive pronoun of the subject was able to occupy the position of the Object of the verb, and it could be determined that a change in the subject's state occurred as a result of the action. For example, "He is walkiag" would become "He is walkiag himself," in which "himself" would change location as a result of the "walking." Those action utterances which coded movement and resulted in a change of an object’s state, were further subdivided with regard to the type of change that occurred. The change may have been in an object’s attributive state, locative state, or possessive state. Moveaent verbs CodingiaiCaaage in Attribute Staga. The action coded by utterances in this category caused either internal or external changes in the state of affairs of their objects. In the example "The boy is 922128 the cookies," the "eating" results in an external change in the attributive state of the "cookies," whereas "He is teasiag his sister" changes the internal state of the sister. Utterances of this nature were consistent with the statement "subject is verbing SOMETHING." Movament verba_goding_aa§aaage in Locativeggtaga. The action coded by the verbs within this category caused a change in the location or position of their objects (or subjects if no Objects 45 were required). In the example "Ea: the ball down," the "ball" changes location as a result of the "putting." In the example "The girl is swimmiag," the act of "swimming" results in a change in the location of the "girl." These utterances were consistent with the statement "Subject is Verbing (Object/ Reflective Pronoun of the subject} SOMEWHERE." Movement Verbs Coding a Change in Possessive Staga. The action coded by the verbs in this category caused a change in the temporary ownership or possession of an object. The change may have occurred from one person to another, as in the example "giga me the candy," or from.a location to a person, as in the example "I EQEEEL the dress." These utterances were consistent with the statement "Subject is Verbing Object {to/for} WE." The actions in this category were distinguished from actions resulting in a locative change by the fact that these actions specifically result in the object becoming newly located into the hands of a person or animate being, rather than becoming located to a new position in space. Action verbs Codiqg_§9nchaage of State. The action verbs in this category did not specify a change of an object’s state and could occur within both movement and nonmovement contexts. In the movanent example "She is Elam house" and the nonmovement example "She is holdiag the ball," the "playing" and "holding" do not result in any inherent changes in the state of an Object. Although various changes may occur as these actions are carried out, specific changes were not identified by the verb itself. 46 Those action verbs which coded nonmovement, and did not result in a specified change of an object’s state, were further classified according to the type of unchanging Object state which they described. For verbs which did not require an object, the action described an unchanging state of the subject. The unchanging state of affairs may have been in the external state, positional state, or possessive state of the Object. Nonmovement verba Coding a Nonchaage in External Spaga. The verbs in this category referred to internal or mental processes. For example, "The boy is thinkiag" describes a conscious mental activity that the "boy" does. Egaagvement Verbs Coding a Nonchange in Poeitioaal Staga. The action verbs in this category described the maintenance of the positions or locations of objects. For example, "sitting" within the utterance, "The boy is sittiag on the floor," describes the position of the "boy." Nonmovagent Verbs Coding a Noncaaage in Poasessive State. The action verbs in this category described the maintenance of a temporary ownership or possession of an Object. FOr example, "holding" in the utterance, "I am holdiag the doll," describes the placement of an Object in the hands of a person or animate being. Summagy of subcategory Fraaewora The combination of each of these aspects of movement and change resulted in the differentiation of action into eight subcategories, as outlined in Figure 1. Thus, each action utterance was assigned 47 to one of the following subcategories, on the basis of its movement and change characteristics: 1) 2) 3) 4) 5) 6) 7) 8) Movement-Change of Attributive State Movement-Change of Locative State Movement-Change of Possessive State Movement-Nonchange of State Nonmovement-Change of State Nonmovement-Nonchange of External State Nonmovement-Nonchange of Positional State Nonmovement-Nonchange of Possess ive State m .oocuzu one unoeo>oz uo caduouuoucu as» com: oommm coduo< uo noauooouoonsm .a oasmwa Page” 38.5» acefia ontfifi 5H2 was! 9:an 9:53 8 35m oumum 4 gang Hgdu 9:5..ng onstage eeefizzu EBB? moflxw -8... no -38 to «actuate. no mecflhflaz mEGthaz P P ouuum uo omcucucoz _ 8:25 b unusu>oscoz _ ouaum u» accuse macaw no owconocoz mndwaun .Fnlpflfl. DGfimlo 9.13 9:23 $38 mantra masses unseen .wa>fim unusaxa sushi: 35m 35m 938m 33m 93 .58 so 933 L552 nFmguebiflwa£u eomJ15n. ouaum.uo ounces _ acoeo>oz _ 49 A formal procedure was developed to determine (1) the extent to which this investigator’s assignment of categories concurred with others and (2) the extent to which the observers independently agreed in their assignment outcomes. Observer Agreement With the Subcategoriz_al;ion of Action Verbs Three graduate students were asked to participate as judges in the determination of observer agreement in subcategorizing action verbs . Each student was provided with instructions regarding subcategorization procedures (Appendix B), a copy of the utterances to be categorized (Appendix C), and a copy of the contextual notes for each utterance. Preparation of Task for Observer Judgment Fifty-four verbs were randomly selected from the set of 1,627 action utterances coded. The nunber of action verbs chosen from each of the eight categories varied as shown below. Movement-Change in Attributive State: 10 Movement—Change in Locative State: 10 Movement-Change in Possessive State: 5 Movement-Nonchange of State: 10 Nonmovement-Change of State: 5 Nonmovement-Nonchange of External State: 5 Nonmovement-Nonchange of Positional State: 5 Nonmovement-Nonchange of Possessive State: 4 50 Ten utterances were chosen from those subcategories that had a greater nunber of action verbs represented, and five utterances were selected from those subcategories with fewer action utterances. Only four action verbs were taken from the category Nonmovement- Nonchange of Possessive State because there were only four verbs represented by the subjects in this category. The total of 54 utterances was judged to be a reasonable amount for the time allocated to the task. Each action verb was randomly selected across all verbs within each category until the quota for that category was met. To ensure the selection of a variety of action verbs, no verb was selected twice within one category. Once the verbs were chosen, the specific utterances within which the verbs occurred, were randomly selected across all subjects. The utterances were then transcribed in their original form onto a separate reliability form, in random order. The original transcription forms were photocopied so that the assigned semantic subcategories could be deleted. The original utterances and context notes were clearly numbered to correspond to each action utterance on the reliability forms. In this way, the judges had access to the situational context to aid their designation of action subcategories. Description of Jaggaa The three graduate students who participated as judges were potential professionals in the field of speech and language 51 pathology. The judges were blind observers, and they individually received oral and written instructions of the categorization task. All judgments took place in the language Sciences Laboratory within one week’s time period. PreaanLation of the Utterances to Judgg In order to reduce the nunber of instructions the judges needed to attend to at one time, the categorization of utterances was introduced in four phases presented in the same sitting. The judges were first required to distinguish between those action verbs that referred to movement and those that did not. Following these judgments, they reviewed the same utterances to determine those action verbs that code a change in object state and those that do not. In the third phase, the judges were presented with movement verbs coding a change of object state. They were required to determine whether the action resulted in an attributive, locative, or possessive change in the object’ 8 state. In the final phase of the observer agreement study, the judges were required to differentiate among nonmovement action verbs that referred to a nonchange of object state. They were required to determine whether the action described a nonchange of external, positional, or possessive state . Percentage of Obaerver Agreement The percentage of those utterances which were subcategorized in agreement with this investigator was calculated for the judgments 52 between Movement and Nonmovement categories , Change and Nonchange categories, type of Change -— Locative, Attributive, and Possessive categories, and type of Nonchange -- Positional, External, and Possessive categories. As Table 1 indicates, the average inter- judge agreement for each of these characteristics ranged from 94- 100%, with an overall average of 95%. It was concluded, therefore, that this investigator’s subcategory assignments had some measure of concurrent validity. Table 1 also shows that there was very little variation among the subjects for any single set of categories judge. Note that the largest observed difference occurred within the Movement/Nonmovement judgnents, which ranged from 91-98%. It was therefore concluded that there was also some measure of inter- observer reliability in the assignment of subcategories. Table 1. Results of Inter-judge Reliability Measures. Reliability Judgments Subjects Average 31 S: S! Movelnent/Nomnovement 91% 93% 98% 94% Change/Nonchange 91% 93% 93% 92% Change- Locative/Attributive/ Possessive 92% 92% 96% ' 93% Nonchange- Positional/External] Possessive ' 100% 100% 100% 100% 53 ggpilation of Refllts The data were tabulated.to determine: 1. The percentage of action utterances subcategorized -- calculated by dividing the nunber of action utterances which were subcategorized by the total nunber of action utterances . 2. The nunber and type of subcategories of action utterances represented -- determined by scanning the data collection forms for each child and noting which subcategories have been represented at least once by the child. 3. The diversity of the verb forms within each subcategory of action - determined by calculating the percentage of different verb forms within each category. The Wilcoxan matchedepairs signed-ranks test and the Friedman Two-Way ANOVA by Ranks were applied to the data, as appropriate. Outcomes Which Would Sapport the Existence of Subcategories of Action Utterances in Normal lLDevelopingL Children It was expected that at least 90% of the children’s action utterances would be represented by the subcategories of action outlined in Figure 1. It was also expected that, while the categories of verb forms would be similar across the children, there would be individual variation with respect to the inventory of specific verb forms used within the subcategories. 54 Part 2 - Action Subcategories in Clinical and Normal Samples The following section outlines the procedures designed to answer the second research question: Can the application of identified action subcategories reveal differences between normal and language-impaired children’s action utterances? The subjects were selected from an existing data base that included clinical and normally developing children between the ages of 3:0 to 8:0 years of age, and provided audio-visual records of spoken language samples. Unlike the data base used in Part 1, these samples were collected in a University Laboratory setting, rather than in the home. Description of Subject Characteristics The subjects consisted of two language-impaired and two non- language-impaired children matched on the basis of age, sex, ethnicity, and socio-economic background. All four children were white males from working-class families. The first of the language-impaired subjects (M.W.) was 4:5 years of age with normal nonverbal cognitive abilities based on a Laigag age equivalency of at least 5:6 years of age. His MLU of 3.44 placed.him in Brown’s Early Stage Iv with a predicted age of approximately 3:2 years. .M.W.’s scores on the N§§I placed his receptive performance at the 25th percentile and his expressive performance less than the 10th percentile. These results suggested that M.W. exhibited a significant expressive language delay of approximately one year compared to his chronological age. M.W. had At 55 been receiving speech and language therapy for approximately one year through his preschool program. He was very cooperative and talkative throughout the testing procedures . A normally developing child (D.B.), 4:6 years of age, was selected as a match to M.W. . He was also very outgoing, and animated throughout the testing procedures. D.B. demonstrated nornal nonverbal cognitive skills based on a _L_ei_t_eg age equivalency of at least 5:6 years of age. His MLU of 5.37 placed him in Brown’s Stage v with a predicted age equivalency of approximately 4:6 years of age. The results of the ELSE revealed that D.B. was performing receptively at the 75th percentile and expressively between the 50th and 75th percentile. These results suggested that D.B. ’s receptive and expressive language skills were within normal limits for his age. D.B. was attending a preschool at the time of this study, but had never received speech and language therapy. The second of the language-impaired children (J.K.) was 3:10 years of age . He also demonstrated normal nonverbal cognitive abilities based on a M age equivalency of at least 3:10 years of age. His MLU of 3.76 placed him within Brown’s late Stage IV- Early Stage v, with a predicted age equivalency of 3:4 years. Administration of the m was attempted, but J.K. would not cooperate with the testing procedures. While these results did not provide strong support for the diagnosis of language-impaired, J.K. was identified as language-impaired, based on a December 5, 1986 assessment, by the speech pathologist at the preschool he attended. At that time his spontaneous speech consisted primarily of one-word PE 0h; 57 throughout the language sample collected in the Language Sciences Laboratory. A summary of the subject characteristics is outlined in Table 2. subjects selected to participate in this study were free of frank neurological insult and physical, sensory, or motor disability, as determined by clinical records and parental reports. In addition, the children were included if they received a normal range Leiter IQ score. The Leiter International Performance Scale (LIES) was chosen to test for intelligence because it was assumed that language-impairment would not penalize the child’s performance. In addition, it enabled the subjects of this study to be more comparable with the other research. The LIES required children to match one set of blocks to another on the basis of perceptual and conceptual information. No verbal instructions were provided to the children. The scoring procedures involved basal and ceiling levels with each level corresponding to an age-equivalency score. The language-impaired subjects were selected on the basis of percentile scores of 10 or lower, relative to the child’s age group on one or both of the Northwestern SyntaagScreeniag:Test4(NSST) subtests. The nonimpaired group at the same age were included if they received percentile scores of 50 or higher for both N§§I subtests relative to the child’s age. The fl§§I has a receptive and an expressive subtest consisting of eleven items each. The receptive portion required the child to point to one of four pictures representing the utterance spoken by the examiner. The child was required to provide a delayed imitative response in the 58 Table 2. Subject Characteristics Characteristics Subjects M.W. D.B. J.K. A.D. Language- Normal Language- Normal Classification Impaired Impaired Age 4:5 4:6 3:10 3:11 Laigag Age Equivalency ‘35:6 35:6 33;10 34:8 MLU Score 3.44 5.37 3.76 7.37 Brown’s Stage Ear.St.IV Pst.St.v Late St.IV Pst.St.V Predicted Age 3:2 4:6 3:4 >4;10 N§§I Receptive Percentile 25 75 -- <10 Expressive Percentile <10 50—75 -- >90 Testing Behavior animated animated uncooperative shy 59 expressive portion of the NSSI. Each correct response received a score of one point, and the total score was tabulated for each subtest and compared to statistical norms to calculate a percentile ranking. In addition, the language-impaired children had been previously diagnosed, and were receiving therapy. On the other hand, the normally developing children were not receiving therapy and had never been suspected of having language problems. Effort was made to select children from a common socio— economic background, as indexed by parental occupation and predictions about co-occurring incomes. They were chosen from the Head Start Program and elementary school populations of Lansing, Michigan. This plan was approved for use by human subjects by the Michigan State University Coulnittee on Hanan Subjects (See Appendix D). Subject Selection Procedures Screeniag Procedures The subjects were selected in two phases of screening activity. The first phase identified the clinical and nonclinical children who potentially met subject selection criteria. The second screening phase verified.whether the subject selection criteria.had been met. Phase 1 of Screening Activity. The first phase of the screening activity involved the classroom teachers and speech- language pathologists in the school programs. They completed brief questionnaires on every child in the age ranges of interest (see 60 Appendix E). The teachers provided information about children who had never been referred to speech therapy or special education classes whereas the speech pathologist provided information about the language-impaired children. The questionnaires required yes/no responses to questions regarding demographic features (e.g., age, sex, educational placement, health, history, and socio-economic status), professional judgments, and access to the child’s school records. In addition, it was determined whether scores were available for the Bag: and the nga. Further screening of those children who potentially met sUbject selection criteria was completed as the questionnaires were returned. The children’s identity remained anonymous until they were selected as potential subjects. Phaseag of Screeningaactivity. The second phase of screening involved the administration of the fl§§I and LIES to determine if the potential subjects met criteria for selection. In addition, a fifty utterance language sample was collected for the calculation of a mean length of utterance. Informal Observations of the oral peripheral structure and functioning, and verbal and nonverbal behavior such as speech intelligibility, vocal attributes, and response to verbal commands were noted to ascertain body integrity. A standard checklist was used for the purpose of applying the same observations to each child screened (see Appendix F). Those children who did not meet subject selection criteria were not Observed further. 61 The testing during this phase was conducted in the child’s home environment after written parental consent was Obtained (see Appendix G). The language-impaired children were selected first, and then normal children of comparable age, sex, ethnicity, and socio- economic background, were sought. Procedures Used to Obtain aaagaage Samples The children’s spoken language was recorded using audio-visual equipment that consisted of a portable color camera (Sony BMC-660) equipped with view finder and automatic light control and self- contained video-cassette recorder supplemented by a microphone (Sony ECM-ZZOO). The sample also was audio-recorded using a cassette recorder (Sony TC-205) with a tie-tack microphone (EMC- 150T) attached to the child’s clothing at the collar. All Observations took place in the Language Sciences Laboratory at Michigan State University during the hours of 10:00 to 4:00. Prior to the language sample, the subjects had participated in a structured locative task designed for another study. Following this structured task which lasted 30-40 minutes, the children were given a short break and then engaged in a play activity with the investigator, so that a 50-60 minute language sample could be collected. At this point, the children were usually eager to play and were familiar with the investigator who also had administered the structured task. 62 A core set of toys was used consistently across all subjects, and centered around four themes, each being the focus of a 10-15 minute play interaction. The first theme introduced objects for a picnic scene. The child played with objects that included a ‘mommy, a daddy, a son, a daughter, a dog, a picnic table, a picnic blanket, flowers, plates, knives, napkins, raisins, and crackers. The second theme revolved around modes of transportation. The objects included a jeep, a helicopter, a plane, boats, and horses. The third theme centered around animals such as monkeys, horses, cows, sheep, roosters, pigs, and a barn. The final theme centered around machinery that included a dunp truck, tractor and wagons that were handled in relation to rocks, stones, and lichens. The first two activities took place on a landscape set depicting a hill with a tunnel through it, a river with a bridge over it, and a grassy field. The third and fourth activities took place on a green mat depicting grass and placed on the floor. These toys were selected because they allowed the children to perform a wide variety of actions and appeared to be of high interest value to children. The play activity involved the child and the investigator, with the investigator imposing a.minimum amount of structure except for controlling the type and sequence of toys that were available for the child’s play. All the toys were kept out of the child’s sight, in separate boxes according to the theme with which the toy was associated. The investigator generally constructed each activity for the child and then allowed the child to position the toys on the set and talk 63 about what they were doing. The themes were introduced, one-at-a- time, in a consistent order as previously described. As the child appeared to lose interest in an activity or reduced the amount of verbal output, a new toy was introduced, and the other toys removed. The child was allowed to play with toys in whatever manner he wished. The investigator attempted to keep the language-sample as child-oriented as possible by engaging in parallel play and asking open-ended questions. In instances when the child was not very talkative, or was very repetitious in his/her play, the investigator became more directive. Form of the Data The data were available in audio-visual form on color Beta video cassettes. The audio-visual record provided the situational and linguistic context in which the child’s action utterances were produced” This was helpful in interpreting the underlying meanings of the child’s utterances. The audio-visual recording was viewed and all utterances were transcribed orthographically on the standard forms used in Part 1 of this study. subject identification, utterance numbers, counter numbers, and general context were also noted on these forms. Procedurea for Identifying Action Utterances The utterances coding action were identified in the same manner as in Part 1 of this study. The audio-visual tape was reviewed in conjunction with the written form. The investigator looked 64 primarily at the verb in each utterance to identify those coding actions on the basis of the operational definition. It was also noted whether the verb was used correctly within the situational and communicative context. Procedures for Assigning subcategories of Action Utterances The contexts were considered in order to infer the specific meanings encoded within each action utterance. The utterances were then assigned to the subcategories developed in Part 1 of this study, based upon the movement and change characteristics of the action. Tabulation of the Data The same analyses performed on the original norming group, were performed on the data collected from the normal and language- impaired children in Part 2 of this study. The Rank Suns test was applied to the data, as appropriate, to determine if significant differences existed between groups. Outcomaa Which WOuld Sapport the Aggament of Differences in Semantic Relations Used by Noraal and Clinical Groups It was expected that all of the children would be coding action in some form. However, it was also expected that differences would exist in any one of the three aspects of data tabulated. The clinical children were expected to code fewer of the action subcategories than the normal children. In those subcategories of action which the clinical and normal children had in common, it was 65 expected that the clinical children would represent less diversity in their verb forms. For example, the clinical child might use only two verb forms to code a particular subcategory of action, whereas the normal child might use five different forms within the same subcategory . RESULTS Part 1 - Action subcategories in a Normative Sample The purpose of this section is to describe the distribution of action verb subcategories in the normative sample. Ninety-six percent of the action utterances produced by the children in this study were accounted for by the following eight subcategories: Movement-Change of Attributive State Movement-Change of Locative State MOvement-Change of Possessive State Movement-Nonchange of State Nonmovement—Change of State Nonmovement-Nonchange of External State Nonmovement-Nonchange of POSitional State NOnmovement-Nonchange of Possessive State Although all categories were represented in the language samples of one or more subjects, they did not occur with equal frequency, nor did they include the same nunber of verb types. In the following sections, 1) the relative frequency distribution of subcategories and 2) the lexical composition of each category are described for each subject. 66 67 The Regive Frequency Diatribution of Verb Subcategories The proportion of different action verbs represented in each of eight verb categories was calculated for each subject. These proportions are displayed in Table 3 and the mean proportions are graphically sunnarized in Figure 2 for visual inspection (see Appendix H for complete data record). Inspection of Table 3 and Figure 2 reveals that by far the greatest number of action utterances were represented by the movement categories which accounted for more than 3/4 of the data. The Movement—Change of State categories alone accounted for approximately 2/3 of the different action verbs in this study. In an attempt to make the relationships between categories more apparent, the information presented in Table 4 has been subdivided into .aller tables which show percentage values for selected sets of categories, as will be described in the following sections. W of MovementZNonmovement Categories The data in Table 4 collapses the categories into Movement and Nonnovement, making it apparent that a greater diversity of action verbs are characterized by movement rather than nonmovement . The Wilcoxan matched-pairs signed ranks test (Orkin and Drogin, 1974) was applied to these data and revealed a significant difference (W = 36, P (W _>_ 20) = .098) between the categories of Movement and Nonmovement. 68 Table 3. The Percentage of Different Action verbs In Each Category For Each Child in Normative Sample. Subcategories Subjects X Sd KM ST DW DD MW LA CW EC N (177)(109)(274)(205)(224)(199)(256)(120) Movement- Change Locative 26.3 21.9 27.8 32.0 23.2 27.8 41.0 36.8 29.6 2.5 of State Attrib. 23.7 25.0 24.1 28.0 32.1 24.1 21.3 21.1 24.9 1.4 Possess. 10.5 15.6 11.1 6.0 5.4 13.0 9.8 10.5 9.5 1.3 subtotal 60.5 62.5 63.0 66.0 60.7 64.9 72.1 68.4 64.8 1.5 Movement- Nonchange 31.6 15.6 18.5 14.0 25.0 14.8 9.8 21.1 18.8 2.6 of State Nonmovement- Change of 2.6 9.4 5.6 10.0 7.1 9.3 1.6 7.9 6.7 1.2 State Nonmovement- Nonchange Pesition 2.6 6.3 5.6 6.0 1.8 5.6 6.6 2.6 4.6 .7 of State Internal 0 6.3 1.9 4.0 1.8 3.7 8.2 0 3.2 1.1 Pessess. 2.6 0 5.6 0 3.6 1.9 1.6 0 1.9 .8 subtotal 5.2 12.6 13.1 10.0 7.2 11.2 16.4 2.6 9.8 1.7 Movement—Change of Locative State IMovement-Change of Attributive State Movement-Change of Pessessive State .Movement-Nonchange of State Nonmovement—Change of State Monmovement- Nonchange of Positional State Nonmovement- Nonchange of External State Nonmovement- Nonchange of Pessessive State 69 27.1 29.6 32.1 23.5 24.9 26.3 ! | 8.2 9.5 10.8 I 16.2 18.8 21.4 I 5.5 6.7 7.9 I 3.9 4.6 5.3 I (2.1 3.2 4.3 F"1r“T-w p._4u.L~J I I1.1 1.9 2.7 tvmnrry pres—.3 I 0 5 10 15 20 25 30 X Percentage of Different Action Verbs 1,1 s.d. Figure 2. The Mean Percentage of Different Action Verbs 1_1 Standard Deviation Within Each Category. ‘ 70 Table 4. The Percentage of Different Action Verbs In the Movement and Nonmovement Categories . Subcategories Subjects X Sd KM ST DW DD MW LA CW EC Movement 92.1 78.1 81.5 80.0 85.7 79.7 81.9 89.5 83.6 1.9 Nonmovement 7.8 22.0 18.7 20.0 14.3 20.5 18.0 10.5 16.5 1.9 Comparison of Change/Nonchange Categories Table 5 sunnarizes the diversity of action utterances within the subcategories involving a Change of State and Nonchange of State revealing a substantial advantage of the Change of State categories over the Nonchange of State categories in representing the action verbs used by subjects. Application of the Wilcoxan matched-pairs signed- ranks test indicated that this difference was statistically significant (W = 36, P(W ; 20) : .098). However, close inspection of Table 5 reveals that the interaction of movement with change contributed greatly to this difference. When the Wilcoxan matched-pairs sign-ranks test was applied to the subcategories of Change of State and Nonchange of State separately for the Movement and Nomnovement subcategories , as in Table 6 and Table 7 , it becomes apparent that the dominance of Change of State over Nonchange of State is characteristic only within the Movement categories (W = 36, P(W > 20) = .098). Within the Nonmovement category, the opposite difference was noted, with the Nonmovement- 71 Table 5. The Percentage of Different Action Verbs In the Change of State and Nonchange of State Categories Irrespective of Movement or Nonmovement Classification. subcategories Subject X Sd KM ST DW DD MW LA CW EC Change of State 63.1 71.9 68.6 76.0 67.8 74.2 73.7 76.3 71.5 1.7 Nonchange of State 36.8 28.2 31.6 24.0 32.2 26.0 26.2 23.7 28.6 1.7 Table 6. The Percentage of Different Movement Action Verbs In the Change of State and Nonchange of State Categories. subcategories Subjects X Sd KM ST DW DD MW LA CW EC Movement— Change of 60.5 62.5 63.0 66.0 60.7 64.9 72.1 68.4 64.8 1.5 State Mbyement- Nonchange of State 31.6 15.6 18.5 14.0 25.0 14.8 9.8 21.1 18.8 2.6 72 Table 7. The Percentage of Different Nonmovement Action Verbs In the Change of State and Nonchange of State Categories. subcategories subjects X Sd KM ST DW DD MW LA CW EC Nonmovement- Change of 2.6 9.4 5.6 10.0 7.1 9.3 1.6 7.9 6.7 1.2 State NOnmovement- Nonchange of State 5.2 12.6 13.1 10.0 7.2 11.2 16.4 2.6 9.8 1.7 Nonchange of State category representing a significantly greater percentage of action verbs than the Change of State category (W = -23, P(W 2.20) = .098). ggnparison of Change of Locative/Attributive/Possessive State Categories The subdivision of the Movement-Change of State category into the type of change -- Locative, Attributive and Possessive -- is represented in Table 8. The Friedman Two-Way'ANOVA by Ranks (Orkin & Drogin, 1974) revealed a significant difference among the categories Movement-Change of Attributive, Locative, and Pessessive States (Xr’ (2) = 13, p ( .005). The Wilcoxan matched-pairs signederanks test was then applied to compare each pair of categories. The results indicated that Movement-Change of Locative State represented a significantly larger percentage of action verbs than did the Movement-Change of Attributive State (W = 20, P(W 2.20) = .098) or the Movement-Change of 73 Table 8. The Percentage of Different Movement Action verbs In the Change of Locative, Attributive and Possessive State Categories. subcategories subject X Sd KM ST DW DD MW LA CW EC Change of Locative 26.3 21.9 27.8 32.0 23.2 27.8 41.0 36.8 29.6 2.5 State Change of Attributive 23.7 25.0 24.1 28.0 32.1 24.1 21.3 21.1 24.9 1.4 State Change of Possessive 10.5 15.6 11.1 6.0 5.4 13.0 9.8 10.5 9.5 1.3 State Possessive State (W = 36, P(W 2_20) = .098). The Movement-Change of Attributive State represented a significantly larger percentage of action verbs than did the Movement-Change of Possessive State category (W : 36, P(W 2_20) = .098). Qannaniaon of Nonchange of Pesitional[External/Possessive State Categories The Friedman Two-way ANOVA by Ranks was applied to the Nonmovement- Nonchange of State categories, represented in Table 9, significant differences were not revealed between the categories of Nonchange of Pesitional, External, or Possessive State (Xra (2) = 3.1, p g_.05). 74 Table 9. The Percentage of Different Nonmovement Action Verbs In The Nonchange of Positional, External and Pessessive State Categories. subcategories subjects X Sd Nonchange of Pesitional 2.6 6.3 5.6 6.0 1.8 5.6 6.6 2.6 4.6 .7 State Nonchange of External 0 6.3 1.9 4.0 1.8 3.7 8.2 0 3.2 1.1 State NOnchange of Pessessive 2.6 0 5.6 0 3.6 1.9 1.6 0 1.9 .8 State The Divepaity of Verb Forns Within Each sancategory The total population of action verbs within each category varied considerably across subjects. The verbs in Tables 10 through 17 are arranged in order from those represented by all subjects to those represented by only one subject. Lines have been drawn to highlight those verbs used by all subjects and those used by only one subject. Inspection of these tables reveals that some verbs were produced by a larger group of children than other verbs. Approximately 50% of the verbs within each category were used by only one subject. Fer any one category, no:more than four verbs were used by all subjects. No verb within the nonmovement categories was produced by all subjects. The total nunber of verbs used by all subjects ranged from 4 to 51, across categories. 75 Table 10. The Diversity of Action Verbs In the Movement-Change of Locative State Category. Verbs E: a E 95 2 8 '2' E 2 '8 put 80 come fiat take fall move knock (down) turn run sit 4» push stand + pick + + pull + + junp + bring + throw + + walk + drop + stick + lay + pop (up) + swing spit + ride + drive + backing up + set (back) + carry + follow " + scoot + hit (ball) + spill 1» race + flip + land + leave + fly + swim + roll + +++++ ++++++ + ++++ + ++++++ +++++++ +++++ 4. 4. ++++++++++ + + +++++++++++++ 4. + +++++ ++++ 4. park call (back) + chase + 76 Table 11. The Diversity of Action verbs In the Movement-Change of Attributive State Category. verbs subjects KM ST DW DD MW LA CW EC eat + make open + bite cook break + shake fight + mess + paint close + + wash scare + + pep (popcorn) + + fix + + turn (light on) + + stir + + hit + + kill + out + clean + wrap + take (a bath) + bust + drink + whip + waste + choke + feed + put (together) .4» polish + beat + POP (your hand with elastic) + tease + get (ready) + scratch + tear + cover + wipe + fill + smoke + tie + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + 4. 4. Table 11, contd. 77 Verbs ST subjects DW DD MW LA crack fall (apart) mix bathe knock (you out) lock mock (tease) teach hurt Table 12. Pessessive State Category. The Diversity of Action Verbs In the Movement-Change of Verbs ST Subjects DW DD MW LA CW get gave take find buy lose catch roll bring see (give it) leave pay pick steal + + + + + + + + + + + + \‘er 78 Table 13. The Diversity of Action Verbs In the Movement-Nonchange of State Category. verbs Subjects E 59: 3 DD MW LA CW EC Play 883’ look tell show + help call laugh cry + + messing (fooling) + + talk read + + holler + + knock (on door) + cheat + check (heartbeat) + rub + dance + take (a picture) + ask + pee + sing + feel + peep + taste + reach + use + work + kiss + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + 4. 4. + 79 Table 14. The Diversity of Action Verbs In the Nonmovement-Change of State Category. Verbs Subjects KM ST DW DD MM' LA CW EC stop + + + + + + get (001d) + + 4. die + + go to sleep + + finish + + burn + + learn + freeze + make (me sick) + (school) open + come (on TV) + Calling (M137) 4. win + boil + Table 15. The Diversity of Action Verbs In the Nonmovement-Nonchange of Pesitional State Category. verbs Subjects KM ST DW DD MW LA CW EC wait + + + + stay + + + + ait + + + lay + hold still + stand + hold (on) + step + hide + 80 Table 16. The Diversity of Action Verbs In the Nonmovement-Nonchange of External State Category Verbs Subjects KM ST DW DD MW LA CW EC watch + + + + dream + + think + + bet + + ignore + sleeping + remember + Table 17. The Diversity of Action verbs In the Nonmovement-Nonchange of Pessessive State Category. Verbs Subjects KM ST DW DD MW LA CW EC hold + + + get (i.e. keep) + + save + + Keep + 81 The overall distribution of verbs reveals that all subjects used at least one verb in each of the categories except the NOnmovement-Nonchange of Internal State and Nonmovement-Nonchange of Possessive State categories. This distribution suggests that some categories (Movement- Change of Locative State, Movement-Change of Attributive State, Movement- Change of Pessessive State, Movement-Nonchange of State) are more dominant than other categories (Nonmovement-Nonchange of Internal State, Nonmovement-Nonchange of Possessive State). Part 2 - Action Subcategories in Clinical and Normal Samples The purpose of this section is to compare the distribution of subcategories in the normal and language-impaired groups. Based on the results presented in Part 1 of this study, it would be expected that the normally developing subjects in Part 2 of this study would demonstrate the following trends in the distribution of their action verbs within the subcategories: 1. The Movement categories would represent a larger percentage of action verbs than the Nonmovement categories. 2. Within the Movement category, the Change of State categories would represent a larger percentage of action verbs than the Nonchange of State categories. 3. Change of Locative State would represent a larger percentage of action verbs than Change of Attributive State which would represent a larger percentage of action verbs than Change of Possessive State. 82 4. Little difference would exist between the percentage of action verbs represented by the Nonchange of Positional, Internal or Pessessive State categories. As with the normative sample, all categories were represented in the language samples of at least one of the two language-impaired and nonimpaired children. Hewever, the categories did not occur with equal frequency, nor did they include the same number of verb types. The following sections describe for each group 1) the distribution of verb subcategories and 2) the lexical composition of each category. The Relative Fregnengy Diatribution of verb subcategories for Normal and Langnage-Impaired Groups The percentage of different action verbs represented by each category was calculated for both the normally developing and language- impaired subjects. These results are displayed in Table 18 and the mean proportions are graphically summarized in Figure 3 for visual inspection (see Appendix I for complete record of this data). The rank order of the subcategories was compared between the normal and language-impaired groups, using the Rank-Sums test (Senders, 1958). No significant difference was found between the groups (T1 = 70, Th = 66, p(T $_49) = .05). The normal and language-impaired subjects demonstrated similar patterns in the distribution of action verbs across the eight subcategories. The infOrmation presented in Table 18 has been subdivided into smaller tables in an effort to determine whether the trends observed in 83 Table 18. A Comparison of the Percentage of Different Action Verbs Within Each Category Between Normal and Language—Impaired subjects. subcategories Normal X Rank Language X Rank subjects Impaired Subjects DB AD MM JK N (238) (42) (182) (93) Movement— Change Locative 42.3 57.1 49.7 1 42.6 46.9 44.8 2 of State Attrib. 26.9 19.0 23.0 4 21.3 25.0 23.2 3 Possess. 3.8 4.8 4.3 10.5 4.3 9.4 6.9 7 subtotal 73.0 80.9 77.0 68.2 81.3 74.9 Movement- NOnchange of State 11.5 9.5 10.5 6 17.0 9.4 13.2 5 Nonmovement- Change of State 1.9 0 1.0 15 0 0 0 16 Movement- Nonchange Position. 3.8 4.8 4.3 10.5 12.8 0 6.4 8 of State External 3.8 0 1.9 14 2.1 3.1 2.6 13 Possess. 5.8 4.8 5.3 9 0 6.3 3.2 12 Subtotal 13.4 9.6 11.5 14.9 9.4 12.2 T; 70 T2 = 66 Movement~Change of Locative State Movement-Change of Attributive State Movement-Change of Pessessive State Movement-Nonchange of State Movement-Change of State Nonmovement-Nonchange of Positional State Nonmovement-Nonchange of Internal State Nonmovement-Nonchange of Pessessive State 84 49.3 j44.8 ,I-"""-III-HT"_HH. H O. C '5"?- 10 20 30 40 X Percentage of Different Action verbs 50 Figure 3. A.Comparison of the Mean Percentage of Different Action Verbs Within Each Category, Between Normal and Language-Impaired Children. 85 Part 1 of this study are also apparent in the data of our age-matched normal and language-impaired subjects. When the categories are collapsed into Movement and Nonmovement, as presented in Table 19, it becomes apparent that no substantial differences exist between the normal and language-impaired children. Fbr both groups, the same tendency as observed in Part 1 of this study was noted. The Movement categories represented a much larger percentage of action verbs than did the Nonmovement categories. The tendency for the Change of State categories to account for a greater percentage of action verbs than the Nonchange of State Categories, was also demonstrated by both the normal and language- impaired subjects, as indicated in Table 20. No substantial differences are noted between the normal and language-impaired children. As noted in Part 1 of this study, the dominance of Change of State hold true only within the Movement categories, as revealed in Table 21. In fact, this trend seems to be reversed within Nonmovement categories. Inspection of Table 22 indicates that, among nonmovement verbs, Nonchange of State represents a considerably larger percentage of action verbs than Change of State. Table 23 summarizes the subdivision of the McVement-Change of State category into the type of change—-Locative, Attributive, and Pessessive. Once again, substantial differences between the normal and language-impaired subjects are not apparent. The rank ordering of these categories from largest to smallest is Change of Locative State, Change of Attributive State, and Change of Possessive State, 86 Table 19. The Percentage of Different Action verbs In the Movement and Nonmovement Categories for Normal and Language-Impaired subjects. subcategories Normal X Language-Impaired X subjects Subjects DB AD MW JK Movement 84.5 90.4 87.5 85.2 90.7 88.0 Nonmovement 15.3 9.6 12.5 14.9 9.4 12.1 Table 20. The Percentage of Different Action Verbs In the Change of State and Nonchange of State Categories for Normal and Language-Impaired subjects, Irrespective of Movement or Nonmovement Classification. subcategories Normal X Language-Impaired X subjects Subjects DB AD MW JK Change of State 74.9 80.9 77.9 68.2 81.3 74.8 Nonchange of State 24.9 19.1 22.0 18.3 25.4 31.9 87 Table 21. The Percentage of Different Movanent Action Verbs In the Change of State and Nonchange of State Categories for Normal and Language-Impaired subjects. Subcategories Norml X language—Impaired X Subjects Subjects DB AD MW JK Movement-Change of State 73.0 80.9 77.0 68.2 81.3 74.9 Movement- Nonchange of State 11.5 9.5 10.5 17.0 9.4 13.2 Table 22. The Percentage of Different Normovement Action Verbs In the Change of State and Nonchange of State Categories for Normal and Language-Impaired Subjects. Subcategories Normal X Language-Impaired X Subjects Subjects DB AD MW JK Nonmovement- Change of State 1.9 0 1.0 0 0 0 Norunovement- Nonchange of State 13.4 9.6 11.5 ‘14.9 9.4 12.2 88 Table 23. The Percentage of Different Movement Action Verbs In the Change of Locative, Attributive, and Possessive States Categories for Normal and Language-Impaired Subjects. Subcategories Normal X Language-Impaired X subjects subjects DB AD MW JK Locative 42.3 57.1 49.7 42.6 46.9 44.8 Attributive 26.9 19.0 23.0 21.3 25.0 23.2 Pessessive 3.8 4.8 4.3 4.3 9.4 6.9 respectively. This ordering is the same as that observed in Part 1 of this study, but with much larger differences noted between Change of Locative State and Change of Attributive State. The Change of Locative State category, for both the normal and language-impaired subjects, represented approximately twice as many different action verbs than did the Change of Attributive State category. As Observed in Part 1 of this study, the Change of Possessive State Category accounted for a considerably smaller percentage of action verbs than the preceding two categories. 'The fourth trend Observed in the data of Part 1 of this study was also upheld by both the normal and language-impaired subjects, as demonstrated in Table 24. Substantial differences between the Nonmovement-Nonchange of Positional, Internal, and Pessessive states were not apparent. In summary, it appears that the general trends in the subcategorization of action data, found in Part 1 of this study, have 89 Table 24. The Percentage of Different Nonmovement Action Verbs In the Nonchange of Positional, External, and Possessive States Categories for Normal and Language-Impaired Subjects. subcategories Normal X Language-Impaired X subjects Subjects DB AD MW JK Positional 3.8 4.8 4.3 12.8 0 6.4 External 3.8 0 1.9 2.1 3.1 2.6 Possessive 5.8 4.8 5.3 0 6.3 3.2 been upheld in Part 2 of this study by both the normal and language- impaired subjects. The only exception noted in Part 2 was the greater percentage of action verbs within the Movement-Change of Locative State category as compared with the Movement-Change of Attributive State category. The Diversity of Verb Forna Within Each subcatagory As in Part 1 of this study, the total population of verbs in each category varied considerably between the normal and language-impaired groups. Tables 25 through 32 document the variety of verbs produced by each group within each subcategory. The verbs are arranged in order from those represented by both groups to those represented by only one group. Lines have been drawn to highlight those verbs used by both groups and those used by only one group. Inspection of these tables reveals that while there is a different inventory of verbs between the 90 Table 25. The Diversity of Action Verbs In the Movement-Change of Locative State Category. verbs Group put go get take jump knock drive fall throw fly bring pick drop sit + + + + + + + + + + + + + + + + + dump walk swim leave stick bump tip limb c_r_aw1 push turn move stand pull park P0P backing + + + + + + + + + + + + + + + + + + + + + + + + + land tramp + + + + + + + + + + + + 91 Table 26. The Diversity of Action verbs In the Movement-Change of Attributive State Category. verbs Group N LI get + + take + + leave + give + pass + 92 Table 27. The Diversity of Action Verbs In the Movement-Change of Attributive State Category. Verbs Group eat Open cut crash put clean seal break close bust fill roll slam hook make hit fire paint get tickle turn (on) explode spread set (table) fit lock poke wake + + + + + + + + + + + + + + + + + + + + + + + + + + + + 93 Table 28. The Diversity of Action Verbs In the Movement-Nonchange of State Category. Verbs Group Play say look cry sins use help tell talk wave Leach + + + + + + + + + + + + + + + Table 29. The Diversity of Action Verbs In the Nonmovement-Change of State Category. verbs Group go to sleep + 94 Table 30. The Diversity of Action Verbs In the Nonmovement-Nonchange of Positional State Category. Verbs Group N LI stay + + ait + wait + stand + hang + lay + hold + Table 31. The Diversity of Action Verbs In the Nonmovement-Nonchange of External State Category. Verbs Group N LI sleep + pretend + watch + 95 Table 32. The Diversity of Action Verbs In the Nonmovement-Nonchange of Possessive State Category verbs Subject N LI keep + + hold + + save + normal and language-impaired groups, it is not clear how these words differ, if at all. The overall distribution of verbs across categories reveals that both groups used at least one verb in each of the categories except the Nonmovement-Change of State category. DISCUSSION Sunnary and Conclusions This study was conducted for two purposes. The first was to develop subcategories of the global category action which could account for the majority of action utterances produced by normally developing children. The second purpose was to apply this system of subcategories to compare the action utterances produced by language— impaired children to those produced by their normally developing peers. The results were expected to expand semantic descriptions of child language and provide insight into the nature of language impairment. The action utterances produced by eight normally developing children between the ages 4:3 and 4:6 years were extracted from spoken language samples approximately one hour in length . These utterances were assigned to one of the following eight subcategories on the basis of the movwent and change characteristics coded by the action verb: Movement-Change of Locative State Movement-Change of Attributive State Movement-Change of Possessive State Movement-Nonchange of State 96 97 Nonmovement—Change of State Nonmovement—Nonchange of Positional State Nonmovement-Nonchange of External State Nonmovement-Nonchange of Possessive State These eight subcategories accounted for 96% of the action utterances produced by the subjects. Those utterances that could not be classified included figurative expressions for which the specific movement and change aspects of action meaning were not readily apparent. The relative distribution frequency of action verbs within these categories revealed four general Observations: 1) Movement categories represented a larger percentage of action verbs than Nonmovement categories. 2) Within the Movement categories, Change of State represented a larger percentage of action verbs than Nonchange of State. Contrastively, within the Nonmovement categories, Nonchange of State represented a larger percentage of action verbs than Change of State. 3) Among MOvement-Change of State verbs, Change of Locative State represented a larger percentage of action verbs than Change of Attributive State, which represented a larger percentage of action verbs than Change of Pessessive State. 4) Among NOnmovement-Nonchange Change of State verbs, distributions among the Nonchange of Positional, External, and Pessessive States were not significantly different from.each other. When this system of subcategories was applied to the action utterances produced by normal and language-impaired subjects of 98 comparable ages, the four general observations were upheld regardless of the grouping. The rank ordering of the eight subcategories was very similar for the normal and language-impaired groups. The population of action verbs comprising each category varied considerably for both the normal subjects in Part 1 of this study, and the nornal and language-impaired subjects in Part 2 of this study. While the normal children used verbs that the language- impaired children did not, and conversely, the language-impaired children used verbs that the normal children did not, it was not clear how these words differed, if at all. C_onparison of the Results with Other Stuiies It appears that only one other study (Huttenlocher et al. , 1983) has systeuetically investigated the action utterances of nornally developing children on the basis of movement and change characteristics. However, it is difficult to compare the Huttenlocher et al. results with those reported in this stu'ly because the definitions of the action subcategories were different. Huttenlocher et al. (1983) classified verbs of movement in terms of characteristic movements (e.g. , walk, wave), and separated these from verbs of change (e.g., get, open). They also analyzed both comprehension and production data. Results of their study suggested 99 that, while children comprehend verbs of movement earlier than verbs of change, they rarely produced verbs of movement to describe their own actions, although they did use verbs of change. Contrastively, this study dealt with only production data and, the subcategories of action were defined in a different way. verbs were defined by the interaction of movement and change characteristics. They were classified as movement if any Observable movement was necessary to carry out the action, irrespective of whether or not they were characteristic movements. Because each action verb was defined on the basis of both movement and change features, the verbs of change in the Huttenlocher et al. (1983) study would have been classified in this study as verbs coding Movement-Change of State. The characteristic movement category of Huttenlocher et al. contains verbs that would be classified as either Movement-Change of Locative State (e.g., walk), or Movement- Nonchange of State (e.g., wave). Irrespective of the differences between this study and that of Huttenlocher et al. (1983), both studies suggest that the characteristics of movement and change dominate the early acquisition of action verbs. While trends in the distribution of action verbs revealed in this study were somewhat similar to the results of Huttenlocher et al. (1983), it is not clear why these trends existed. The possible explanations for the fact that some categories represented a greater variety of action verbs than others, must be explored. 100 Interpretation of the Distribution of Action Verbs In accounting for the four general trends in the distribution of action verbs, one may look to two sources. The first source of explanation may be related to the method of language sampling, employed in this investigation. The second source may be found in theoretical explanations of language development. Although the method has been used frequently within the past ten years, limitations of language sampling are being recognized. One such limitation is that language is situation specific. Consequently, the verbs found to be dominant in this study may have been so because the situation encouraged their use. For example, the dominance of the movement verb categories over the nonmovement verb categories may have resulted from the play situations used to collect the language samples. The majority of these situations involved activities requiring the child’s movement of himself and objects. It is possible that, within other contexts, a different distribution of action verbs across categories would be revealed. Interpretation of the Results Within Theories of Langnage Developnant The literature suggests that many variables have an impact on children’s language development. These include perceptual, semantic, conceptual, and.pra¢matic factors, as well as language input factors. Before proceeding with a discussion of these variables, two notes of caution are warranted. The first is that language development is extremely complex and may be accounted for 101 by an interaction of a number of variables. The second note of caution is that theories of lexical acquisition have dealt primarily with the acquisition of Object words. Therefore, they may not be equally applicable to action verbs, particularly in light of Huttenlocher and Lui’s (1979) claim that action concepts are structurally different from object concepts. While recognizing these two prOblems, the following sections will attempt to identify briefly the perceptual, semantic, and conceptual factors involved in language acquisition, and tentatively draw interesting parallels between these positions and the results of this study. Percepnnal Factors. Clark (1974) suggested that, in learning language meanings, children extract the most salient perceptual characteristics of situational examples named, and assume that is what the word refers to. They will then use these characteristics to identify new referents for the same label. The child using this strategy will often misuse newly acquired.words in a manner which Clark refers to as "overextensions." For example, a child who has determined that the word "dog" means a "furry, four-legged animal" on the basis of perceptual features, many overextend the term "dog" to include a cat, horse, cow, etc. Clark (1974) provided evidence for this position primarily through the analysis of children’s overextension terms in early lexical development. She analyzed the observations made in various diary studies of early meanings children seemed to have for certain words. She discovered that the majority of overextensions used by 102 children seemed to be directly derived from the perceptual features of the Object named. The suggestion that the children derive word meaning from the perceptual features of their referents may help to explain children’s acquisition of words referring to objects or static relational terms. However, as Nelson (1985) pointed out, Clark’s (1974) position does not account for children’s acquisition of words that do not have clear, real-world referents. For example, abstract terms, such as "justice,' or relational terms, such as "all gone," do not have Observable characteristic features which the child can identify for meaning. Thus, perceptual factors do not appear adequate to explain children’s acquisition of the Movement and Change subcategories of action presented in this study. However, Clark’s (1974) position may explain children’s acquisition of the Nonmovement-Nonchange of Positional State subcategory of action. The verbs in the category may be acquired through children’s perception of characteristic positions (e.g., sitting, standing) that remain static. The fact that this category comprises such a small percentage of the action verbs used by the children in this study suggests that static actions are not the dominant action features that children talk about. Semntic Features. Clark (1973) also proposed a semantic feature model to help explain children’s lexical acquisition. She suggested that abstract semantic features are derived from the cannon dimensions among terms within a sanantic domin. These 103 dimensions may include size, movement, or time. The semantic features were defined in terms of contrasts along perceptual dimensions, and could be expressed 1_pairs. However, Clark (1973) did not address the issue of how semantic domains were established in the first place. Because this position is based along perceptual contrasts, it is unable to account for those contrasts within a semantic domain that are not perceptually based. As Nelson (1985) pointed out, the contrast between wild and.domestic animals cannot be made on the basis of perceptual characteristics. Nelson (1985) suggested that Clark’s (1973) semantic feature model does not provide the basis on which lexicons are initially acquired. This model may be more relevant to later language development and the distinctions children make among objects within semantic domains or categories. Clark’s (1973) semantic feature model may have some application to the subcategories of action presented in this study. The action verbs produced by Children were characterized by contrasts between 1 movement and i change. Thus semantic features of the action domain may be regarded as movement and change. While the framework of subcategories of action presented in this study appears to be consistent with Clark’s (1973) semantic feature model, the question as to how'these semantic features were established remains unanswered. Perceptual explanations do not seen able to account for the interaction of movement and change characteristics within action, because these features are not static or Observable. 104 Conceptual Factors. While the theories based on perceptual and semantic variables may account for some aspects of language learning, they appear to be inadequate to explain all the facets of word use and interpretation. Nelson (1985) suggested that children must also know about the uses of Objects as well as their appearance. Nelson (1985, 1982) and Bloom and Lahey (1978) are among several writers who have called attention to the conceptual factors influencing language acquisition. This section will briefly discuss each of their proposals and suggest ways in which conceptual factors may help to explain the acquisition of action verbs. In Nelson’s (1985, 1982) functional core model, she proposed that children construct concepts about familiar things in their world before learning words. It is to these concepts, rather than to perceptual features, that early words are attached. Nelson (1985) maintained that the concepts that children constructed had a functional basis in their own experience. The child’s concepts are formed on the basis of experience with Objects in functional situations, and they include information about actions and reactions of Objects in relation to people, especially including the child himself. Nelson’s (1985, 1982) functional core model viewed the child's knowledge of the world, as functionally organized around what he can do and.what things can do. Thus, when the child begins to learn language, the concepts attached to the new word.would be concepts reflecting this functional organization. Children would match their 105 concepts to the adult’s use of a word within a particular context and then extend the use of this word to other items that matched the previous context. In this way, words are mapped directly onto concepts rather than perceptual characteristics. Nelson (1985) provided as evidence for her functional core model the Observation that children tend to name things that did interesting actions such as "ball," or "dog," rather than static things such as "tree" or "kitchen." The functional core model and other conceptual models specify more than how to identify an example by reference to a perceptual form and, therefore, can be used to interpret utterances that do not refer to static real-world objects, as well as utterances that do refer to easily observable objects. While the functional core model may be used to account for relational terms such as action, Nelson (1985, 1982) has only accounted for the acquisition of object words within this model. The recent work of Nelson (1985, 1982) is strongly allied with Bloom and Lahey’s earlier writings of the acquisition of semantic relations. Bloom and Lahey (1978) duscussed the basis of language acquisition in relation to Piaget’s theory of cognitive development. A fundamental premise of Piaget’s theory is that children form mental schemes for acting through actions and patterns of activity with Objects. The result of this action is that children are able to learn about things in relation to one another and in relation to their own actions. 106 While Bloom and Lahey (1978) account for children’s acquisition of each of their defined semantic relations using Piaget’s model, it is their explanation of the development of action relations that is of most interest. They proposed that action relations develop from the concepts the child acquires in acting on his environment. Specifically, it was suggested that, when Objects move, there is a source of movement, some actor, and an effect of the movement which could be some change in location or other change in the mover or in some other object or person (p. 83). They called attention to the regularities in infant activities that permit them to generalize about certain movements. There are also specific effects of specific movements which the child must detect and mentally represent. Bloom and Lahey (1978) suggested that relational concepts develop from the conceptualization of the regularities among movement, the origin of movement, and the effect of movement. It would appear that conceptual factors may be useful in explaining children’s acquisition of relational terms, and in accounting for the general trends in the distribution of action verbs noted in this investigation. In light of the positions of Nelson (1985: 1982) and Bloom and Lahey (1978) one may hypothesize that, as children conceptualize the regularities in various movements and their effects on Objects, they may begin to categorize these actions according to the type of effect they have on objects (i.e., change of locative, attribute, or possessive state). If the child is actively manipulating objects and observing movement patterns and their resulting effects on objects, then it would seem 107 probable that the child would initially acquire a large number of action words to map onto their movement and change of state concepts. It is possible that the features of movement and change of state may be particularly salient to young children, which would account for the large percentage of action verbs coded within the Movement-Change of State categories. In fact, one would expect the Movement-Change of State categories to develop earlier than Nonmovement-Nonchange of State categories. This review of prominent positions in accounting for lexical acquisition is very superficial and only touches the surface observations made within each position. For additional information regarding language acquisition, it is suggested that the original sources documented in the previous discussion be referred to. Interpretation of the Results of the Conparison of Normal and e-Ingpaired Children No significant differences were noted in the nunber and type of subcategories of action represented in the utterances of normal and language-impaired children. For both groups, the pattern of distribution of action utterances among the categories was similar to that noted by the eight subjects in the normative group, with the majority of utterances occurring in the Movement categories . Thus , like previous studies that used global definitions of semantic relations (Freedman & Carpenter, 1976: Fokes & Konefal, 1981; Leonard, Bolders, & Miller, 1976: Coggins, 1979: and Duchan & Erickson, 1976), this study also revealed no significant differences 108 between normal and language-impaired children, even though a more differentiated description of a semantic category was applied. These results are contrary to what might be hypothesized within the framework of the cognitive/representational hypothesis. If a cognitive deficit is at the base of language-impairment, these deficits should be reflected in the semantic system because of the assumed linkage between cognition and semantics. The fact that significant semantic differences were not found between the language-impaired and normally developing children may be a reflection of two possibilities. One possibility is that semantic deficits are not a characteristic of language impairment. The second possibility is that semantic deficits do exist in language— impaired children, but due to methodological problems, they were not revealed in this study. Before accepting the first possibility, the second must be further considered. Methodological Problena of Snall Sanple Siae Which MaygAccount for the Nonsignificant Results In this study, the major emphasis was on the development of subcategories of action verbs, with the comparison between clinical and normal children pilot work to point to future areas of research. Therefore, due to time restrictions placed on the completion of this study, additional children meeting the requirements for participation could not be found. It was decided to continue with the study with only two children per group. While it wasn’t feasible to get more than two subjects per group, it was recognized that this was a problem. 109 Any population would be expected to vary along a continuum for a given measure. Therefore, there is expected overlap between populations of normal and language-impaired children. As a result of the small sample size, the subjects may have been representative of the area of overlap between normal and language-impaired children. In fact, when one considers the subject characteristics, it appears that this may have been the case. The language-impaired children used in this study may not have demonstrated a severe enough language impairment to distinguish them significantly from the normal children. Although J.K. was reported to have started developing language six months prior to this investigation, his expressive language skills were delayed only six months based on his MLU score. A follow-up interview with his speech and language pathologist indicated that, as of September 1987 (3 months following this study), A.D. had passed all language levels on the Zimmerman Preschool Langnaga_§gala and.was in the process of being dismissed from therapy. Therefore, J.K.’s classification as language-impaired may not have been valid. ‘M.W. was also classified as language-impaired and had been receiving speech and language therapy for one year at the time of the study. Based on M.W.’s MLU and w scores he demonstrated an expressive language delay of 15 months. While this classification appeared valid, M3W.’s therapy program may have confounded the results of this study. A follow-up interview with M.W.’s speech and language pathologist revealed that the primary emphasis in M.W.’s 110 therapy sessions was the verb system. It was interesting to note that the reasoning was because verbs are more difficult for language-impaired children to learn because they are less concrete than objects, suggesting from.her Observation that language-impaired children demonstrate weaknesses in their development of action verbs. Because of the treatment focus on verbs, M.WQ may have developed a greater number and variety of action utterances than a language-impaired child who had never received therapy. One of the normally developing children, A.D., also presented characteristics which may have confounded the results. He was an extremely shy child and had never attended pre-school. During the collection of the language sample in the Language Sciences Lab, A.D. produced very limited output. As a result, it is unlikely that a representative sample of his action utterances was Obtained. The confounding factors indicate that it may be premature to conclude that no differences exist between normal and language- impaired children in their action utterances. Further research using a larger sample size and more definitive criteria for judging language-impaired status is warranted. Implications for Fnture Reaearch This study resulted in the development of a reliable framework for subcategorizing action utterances which can describe the majority of action utterances produced by normally developing children, 4:3 to 4:6 years of age. This system can now be used in developmental research to identify trends in the acquisition of 111 action, and to conduct further research comparing language-impaired and normally developing children in the type of action utterances produce. Because in-depth studies of negation and location have been successful in identifying developmental trends, it is prObable that trends in children’s development of action will also be revealed when a more discriminating system of subcategories is applied. The distribution of verbs Observed in this study predicts the following developmental trends which should be empirically tested: 1) Movement categories will be acquired before Nonmovement categories. 2) Within Movement categories, Change of State categories will be acquired.before Nonchange of State categories. 3) Developmental differences may also be noted among the Movement-Change of Locative State, Attributive State, and Possessive State categories. However, it is difficult to hypothesize, at this point, which of these categories would dominate early lexical acquisition. While the data in this study would suggest that Change in Locative State would.be acquired before Change in Attributive State which, in turn,:might be acquired before Change in Pessessive State, this distribution may shift throughout the course of language development. Early in development, the category of Movement-Change in POssessive State may represent a larger percentage of action verbs than later in development. For example, a large frequency of utterances containing "gimme" are often observed in the spontaneous speech of young children. Investigation of the distribution of 112 action verbs within the Movement-Change of Locative, Attributive, and Possessive States, may yield interesting developmental trends. 4) Categories of Nonmovement—Nonchange of Positional, External, and Pessessive State would account for very few of the action verbs in early language development. These categories are likely to be a characteristic of later language development, even beyond 4:6 years of age. It may be possible that trends in the distribution of action verbs within these three categories may be revealed in research with school-aged children when a greater number of these verbs have been acquired. Once developmental research has identified trends in the acquisition of action subcategories, it would be useful to conduct further research into the action utterances of language-impaired children. It is hypothesized that language-impaired children will lag behind normal children in their acquisition of later-developing categories and might exhibit more restricted lexical composition in each category. The system for the subcategorization of action which was developed in this study appears to be a viable tool to be applied.in additional research into both normal language development and the nature of language-impairment. APPENDIX A SAMPLE TRANSCRIPT m APPENDIXA SWETRANSCRIPI‘FORM 41+er .7114 Its “Moi “JG/av. g crowd fi WWW 1%., tie 1.4541513 he 13...... s. c- p r .11 i1 3., 1x... .1 _ 5} HA.“ (Ii-010 Uklnl. .‘JV EEG: i _ «sq 441 i1 g4 13% arm p .3139? i)! gfi‘yaupv-udh q” *3 EGG-O O I \ r H - p » ‘3‘“41—‘36—9 ‘45“ \ 5.1-“ in! i... (heard 93311..” :xC ZS. it"s—v . at (apt-Du as: g: man-=0 hum—Cub . gt»! 113 9:: 34.5.. J Us! 1.864% “5:52.: #3 "E! a; W.” 2....” a..." a...“ .! E—g APPENDIX B INSTRUCTIONS FOR SUBCATEGORIZATION OF ACTION VERBS APPENDIX.B INSTRUCTIONS FOR SUBCATEGORIZATION OF ACTION VERBS RELIABILITY TESTING Thank you for agreeing to participate in this reliability study of the assignment of action verbs to various subcategories. You will be provided with a list of the utterances to be classified as well as a copy of the utterance and the situational context that was recorded at the time the utterance was produced. The contexts will be clearly nunbered in red, to correspond to each utterance nunber. It is very important that the context be considered in each classification you make. Please attempt to classify all of the utterances into the categories provided. 114 115 PDVDLENTLNOMDVEMENT The first set of utterances will be categorized according to whether or not the action involves an overt or observable movement event. Specific instructions for inclusion within either the movement or normovement category are provided on index cards for your convenience. It is very important that you refer to the context of each utterance, even though you feel you can make the judgment on the basis of the verb alone. The use of some verbs may or may not denote movement , depending upon the contexts in which they are used. For example, "I am sitting," within the context of the child moving from a standing position to a sitting position involves movement. However, within the context of the child remaining in a sitting position, "I am sitting" does not involve any movement. Therefore, consider the contexts carefully to determine if movement actually occurs or was possible to occur within the context of the utterance. (hoe you have conceptualized the action that the child is referring to within the context, ask yourself, "Is some type of movement necessary for this action to occur, or can the action occur with no movement at all?" Disregard the tense of the verb, and concentrate on the action itself. Place an‘M in the classification colunn if the action verb involves movement, and a NM if the utterance does not involve movement. 116 MOVEMENT CATEGORY The action verbs within this category involve an overt or observable movement event. Movement includes a wide range of degrees, from verbs such as "I am running home," which require movement of the entire body, verbs such as "She is cutting the cake," which require movement of an extremity, to verbs such as "Look at the dog" which require smaller movement of the head or facial features. Ask yourself, "Can somebody or>something without moving?" If the answer is m, then the act is considered a Movement verb. NOMDW CATEGORY The action verbs within this category do not involve any overt or observable movement events. "She is sleeping" would be an example of a nonmovement verb, because no observable movement is involved in the act of "sleeping." Although some movement may occur throuighout the act of "sleeping," this movement is not necessary to, and does not characterize, the act of "sleeping" itself. Ask yourself, "Can somebody or something withouth moving?" If the answer is yaa, then the act is considered a Nonmovement verb. 117 CHANGE/NONCHANGE You will now be provided with the same set of utterances which will be categorized according to whether or not the action causes any change of an object’s state. Occasionally, the object may be omitted in a child’s utterance, as in the example, "I am aging the ...." In such cases, one can fill in a nunber of objects such as "food," " pizza," "cookies," etc., to determine if the action would result in a change in the object. If it were included in those instances in which the verb does not take an object, the subject becomes affected as a result of the action. In these cases, the reflexive pronoun of the subject is able to occupy the position of object of the verb. For example, "He is walking" becoues "He is walking himself," in which "he" changes location as a result of the "walking." Specific instructions for inclusion within either the change or nonchange category are provided on index cards for your convenience . Once again, the use of some verbs any or may not denote a change of state in the object involved, depending upon the contexts in which they are used. For example, "I get the candy" denotes a change in omrership of the candy within the context of the child reaching for and taking some candy (i.e., the candy is affected because it is transferred from one owner to another). However, within the context of the child holding some candy in his hand, "I get the candy" does not involve a change in the ownership of the candy, but instead is an announcement of his ownership, much in the same way as "I keep the candy" or "I have the candy." Therefore, it 118 is extremely important to consider the context of the utterance to determine whether a change in the state of the Object actually occurs or was possible to occur within the context. For example, if the child says "I get the candy," and he doesn’t actually end up with the candy in his hand, even though there is a bowl of candy in the room, the intention within the child’s utterance was a change in the possession of the candy (i.e., into his hands) even though this change didn’t actually occur. Within this context, it was possible for a change in the ownership of the candy to take place. In other instances, the child may be looking at a picture or pretending to be in a different situation. In these cases, it is necessary to infer whether a change in an Object’s state is possible to occur depending upon the context of the child’s utterance and the response of others to the utterance. Once you have conceptualized the action that the child is referring to within the context, ask yourself, "Is the primary purpose of this action to cause a specific change in the state of an Object?" Disregard.the tense of the verb, and concentrate on the action itself. Place a C in the classification column if the action verb in the utterance involves a change of state of the Object (or subject if no Object is required) and a NC if the utterance does not involve a change of state. 119 CHANGE CATEGORY The verbs within this category involve actions which cause any change of an Object’s state. This change may be in the Object’s physical or attributive appearance (e.g. Close the door), existence (e.g. Stop that.), internal state (e.g. I learned that at school), location (e.g. I am driving the car), or ownership (e.g. Give me the game). Ask yourself, "Is the primary purpose of to cause a specific change in the physical appearance, internal state, location, or ownership of an Object?" If the answer is yaa, then the act is considered a Change word. Place a C in the classification colunn if the action verb in the utterance causes a change of an Object’s state. NONCHANGE CATEGORY The action verbs within this category do not cause any change of an Object’s state. In the example, "She is playing house," the "playing" does not involve any inherent change in the state of an object. Although various changes may occur as these actions are carried out, specific changes are not identified by the verb itself. Ask yourself, "Is the primary purpose of to cause a specific change in the physical appearance, internal state, location, or ownership of an object?" If the answer is n9, then the act is considered a Nonchange verb. Place an NC in the classification column if the action verb in the utterance does not cause a specified change of an Object’s state. 120 CHANGE ATTRIBUTIVE--LOCATIVE—-POSSESSIVE The utterances that will now be presented all involve actions which result in the current change of the object’s state. In those cases in which the verb does not take an Object, the subject undergoes a change of state as a result of the action. In these instances, the reflexive pronoun of the subject is able to occupy the object position in the utterance. For example, "he is driving to town" becomes "he is driving himself to town." It can be seen that in this utterance, it is "himself" that undergoes a change in location as a result of the "driving." YOur task will be to consider the following utterances within their situational contexts, to determine whether the Object affected by the action undergoes a change in attributive, locative, or possessive state. Specific instructions fer inclusion within each category are provided on index cards. Once again, I urge you to carefully consider the situational context in order to place each action verb within its rightful category. For example, the verb "get" can be placed within each of these categories depending upon the context. For example, "get" results in an attributive change in the utterance, "I am getting messy,‘ a locative change in the utterance "I am getting out of here," and a possessive change in the utterance "I am getting the ball." 121 M _(Z‘IANGE IN A'l'l‘RIBUTIVE STATE The action verbs within this category result in either internal or external changes in the state of affairs of their objects. In the example, "the boy is eatgg‘ the cookies," the "eating" results in an external change in the attributive state of the cookies, whereas "he is hurting the dog" results in a change in the internal state of the dog. Utterances of this nature are consistent with the statement, "Subject is Verbing SCHE'IHING. CHANGE IN LCXEATIVE STATE The action verbs within this category result in a change in the location or position of their objects (or reflexive pronoun of the subject, if no object is required). In the example, "the girl is swimming," the act of "swimning" results in a change in the location of the girl. These utterances are consistent with the statement, "Subject is Verbing {Object/Reflexive Pronoun} SOVIEWHERE." CHANGE IN P(BSESSIVE STATE The action verbs within this category result in a change of an object from one person to another (e.g. Give me the candy), or from a location to a person (e.g. I M the dress). These utterances are consistent with the statement, "Subject is Verbing Object {to/for} SOVIEDNE." The verbs within this category can be distinguished from verbs which result in a locative change, by the fact that these actions specifically result in the object becoming newly located into the hands of a person or animate being, rather than becoming located to a new position in space. 122 NONCHANGE EXTERNAL--POSITIONAL--POSSESSIVE The following utterances involve actions which do not result in the current change of an Object’s state. Instead, they describe an unchanging state of an Object. In cases in which the verb does not take an Object, the action describes an unchanging state of the subject. In these instances, the reflexive pronoun of the subject is able to occupy the Object position in the utterance. For example, "he is thinking" becomes "he is thinking himself," and it becomes evident that "he" does not undergo a change in state as a result of the "thinking." Rather, the action "thinking" is simply describing a conscious mental activity that "he" is doing. YOur test will be to consider these utterances within their situational contexts to determine whether the action verb is describing a nonchange of external state, positional state, or possessive state of affairs. Specific instructions for inclusion within each category are provided on index cards. 123 NONCHANGE or EXTERNAL STATE The action verbs within this category describe ongoing mental activities, and are often thought of as mental processes. Fer example, "sleeping" in the utterance, "The girl is sleeping," describes a mental process in the state of the girl. NONCHANGE OF POSITIONAL78TATE The action verbs within this category describe the position or location of objects. For example, "sitting" in the utterance, "the boy is sitting on the floor," describes the position of the boy. NONCHANGE OF POSSESSIVE STATE The action verbs within this category describe the temporary or permanent ownership of an Object. For example, "holding," in the utterance "I am holding the doll" describes the placement of an object in the hands of a person or animate being. APPENDIX C UI‘TERANCES USED IN THE SUBCATECDRIZATION OF ACTION U'I'I'ERANCES APPENDIX C UTTERANCES USED IN THE SUBCATEGORIZATION OF ACTION UTTERANCES mvaapvr/NoN-mvm I’m not chasing you. flappn this. Maya his hand. I gap all of these. Come and.gpink some. I always apaan Santa Claus house. He annp this in here. §EQB mocking mama. flpia it. WOw, see I fiip down. Kermit, um, Melissa try to paaaa me. §§£ ’em up all on their cages. She iaying back. We was dancing. Thedogknockedthemnoffanthimupintheairandslamned him” boom. I’ll save you one. Holdaatill. I can’t -—. Don’t tear the bag and you buy this batch? 124 128 He nanging around. I papa you house fell. Climbing up here. Now, piaan up this house daughter. A balloon fixing up in the air. YOu pang this one. Take this off and aing with it. Yep, I remember how to do it. I goin’ pang these crackers home too. He’s standing up. I’m gonna aaap these, take ’em home. Sitting down in chair. No. When we pay her back, she don’t even give us nothing. Y’all kids better go to sleep ’cause that’s Santa Claus right now. -emy little brother’s gonna pang ours next weekend. He said, Antoine said, pnap --. Not ’til it ppan up. Let me aaa it. Mom, aging me my cupcake. l aaia, in a cup. m. I tickling her. They'ignpping him. Monkey’s walking in the water. I £§§l inside it. I’m a tell her, give me something. 126 Wam- I tickling her. They ignpping him. Monkey’s walking in the water. I gag; inside it. I’m a .tali her, give me something. crashed up. YOu suppose to stop paapin’ Nita. And pagan it up ’cause might get cold. Oh, this gap dirty. 129 Crashed up. You suppose to stop mapin’ Nita. And m it up ’cause might get cold. ()1, this g_o_t dirty. 130 CHANGE AT'I'RIBUI‘IVE--I.OCATIVE-HBSEBSIVE Kermit, an, Melissa try to _ta_asg me. Climbgg’ up here. I tickling her. No, when we Ey her back, she don’t even give us nothing. ana his hard. Ard pgy_e_r; it up ’cause night get cold. Don’t te_;ai' the bag ard you buy this batch? I’m not chasing you. 55; ’em up all on their cages. The dog knocked the man off ard b\_mpe_d him up in the air and alarmed him, boom. Come ard d_rin_k some. I goin’ gka these crackers home too. Now, _cia_an up this house daughter. 01, this go_t dirty. A balloon i‘iyi_ng up in the air. He said, Antoine said, bu__s; -—. Monkey’s walkgg' in the water. You _ta_k_e_ this one. Ii_e that for me. Let me gag it. Wow, see I, I _ffip down. Crashed up. Now Nita go ard gap me a naflrin. He dt._lnp this in here. 131 NON—0PM EXTERNAL--POSITIONALP-POSSESSIVE I gap all of these. Hold atill, I can’t --. Yep. I remember how to do it. She iaying back. Standing arourd. He nanging around. I always dre_;_am Santa Claus house. I’ll aaya you one. I can aaap ’em. They ignoring him. flappn this. Sitting down in chair. gain it. I bets your house fell. APPENDIX D APPROVAL FOR USE BY HUMAN SUBJECTS APPENDIX D APPRONAL FOR,USE BY HUMAN SUBJECTS MICHIGAN STATE UNIVERSITY umvnsm commute ON arsenal momma onmmusumunsnxmufl m humus-rumor: nmwmc um mam . MST LANSING 0 NICHIGAN 0 (flu-IO“ December I, 1986 Q5 15?, awd‘vt 42g} but5tM5¢29y Dr. Ida J. Stockman “Mmtcts Audiology & Speech Sciences ‘39, 378 Communication Arts Building Dear Dr. Stockman: Subject: Proposal Entitled, "Locative Distinctions of . . Clinical and Normal Children" UCRIHS' review of the above referenced project has now been completed. I am pleased to advise that since the reviewers' comments have been satisfactorily addressed. the conditional approval given by the Committee at its November 3, 1986 meeting has now been changed to full approval. You are reminded that UCRIHS approval is valid for one calendar year. If you plan to continue this project beyond one year, please make provisions for obtaining appropriate UCRIHS approval prior to November°3, 1987. Any changes in procedures involving human subjects must be reviewed by the UCRIHS prior to initiation of the change. UCRIHS must also be notified promptly of any problems (unexpected side effects, complaints, etc.) involving human subjects during the course of the work. Thank you for bringing this project to our attention. If we can be of any future help. please do not hesitate to let us know. Sincerely, .' 43‘1C.’ Henry E. Bredeck, Ph.D. Chairman, UCRIHS HEB/jms 132 APPENDIX E QUESTIONNAIRES (XMPLEI‘ED FOR LANGUAGE- IMPAIRED SUBJHJTS APPENDIX E . .0 f4 .. av hiya $9 .2 . so we 6.. o 0 #0 .se.axL:=a:: Lars men as» Jessi .—:.a=:anu==u use. :6 _.wx 3:. ha 1C1_>css lanaslscbc. . .u—ce .codnzu .lsu unl— scc ole: .=.mu ha 1_—=9 dssa nu—auct. c —L .ouctl use I. a». I GUI—s .alOluvan caseboo I a—ILI. cos 0‘ o‘houou .8350. u. . a:¢lunnn use: an Ilalceanahuoa oust. lie I— as. I con—L .IIO.aeLs ::.~:.30uuus he; gala houses. Oamiusu as! Ios‘_—co ova.ua_ do: 1.3::- an.— 0.:h .s_sou.us Ie—uuo_oa casual. LID doo- QUESTIONAIRES mm FOR IANGUAGE-IMPAIRED SUBJECTS 5. : W153... 13“....” unto—rhhuhnk» 134 mn-un Ill _:3;;H .C‘UICP APPENDIX F OBSERVATION CHECKLIST APPENDIX F OBSERVATION CHECKLIST Ability to fomulate and produce the CV syllable /pA /, /tA /, and /kA./. No extraneous vocalizations present during speech. Points to nose, leg, and chair when asked a "where" question. Absence of drooling. Symetry of the facial features. Symetric mouth retraction. Symetric tongue protrusion. Ability to lateralize tongue and move tongue up and down. No obvious paralysis of muscles. Norml gait . 135 APPENDIX G SUBJECT CDNSENT FORM FOR PARTICIPATION IN THE STUDY APPENDIX G SUBJECT (DNSENT FORM FOR PARTICIPATION IN THE STUDY MICHIGAN STATE UNIVERSITY WUAWYANDMM mum-Woman ”WWHTIMDSGENCBW mmmmc minnow-nan SDIJICT CONSINT PORN FOB PARTICIPATION II STUDY I _ consent for _ parent or guardian child’s name to participate in a study of children’s language development. I understand that the goal of the study is to determine how children at different ages talk about where things are located in space. I understand that my child's participation in the study will require at lehet.two but not more than three observation sessions and that each session will last between one to two hours. I understand further that the first observation session could take place in my hone if I request it. I understand that during the first visit. my child will he given two standardized tests of language and language related skills if these tests have not already been given to him/her. The two tests will be the Leiter International Scale and the Iorthwestern Screening Syntax test. I understand that the second and third observation sessions must take place at Michigan State University in the Department of Audiology I Speech Sciences. During these observations. my child will be videotaped while he/she plays with toys and talks to the clinicians about the toys. I understand further that I may observe every observation session with my child. lach session will be scheduled at my convenience within a one nonth time span and will respect my child's tolerance for the play sessions. I understand further that I will be paid 820 for each observation session and be reimbursed for transportation costs at a rate of .21 cents per mile for the use of my car or reimbursed for the entire cost of taxifare. ' I understand that my child’s participation is voluntary and that consent may be withdrawn at any time without recrimination. I understand that the results of the observations will be used for research purposes and not for educational placement. I understand that the results of observations made on my child will be made available to me and to appropriate school officials at my request. I understand that report of these observations will respect my child's right to privacy by not revealing his/her identity. I can be contacted at the following telephone number to set II? the anoint-cumm.mmm ......... 136 APPENDIXH ACTION VERBS PIDDUCED BY THE NOFMATIVE m verbs tput #30 *come #get take fall move pop (up) sit (down) push stand (up) knock (down) 1 turn run pick Jump bring swing throw spit ride pull walk drop drive backing up set (back) carry stick lay (down) follow scoot hit (ball) spill race flip land leave fly swim roll park call (back) chase APPENDIX H ACTION VERBS PRODUCED BY THE NORMATIVE GROUP PINEWEEHPCHANGE IN LOCATIVE STATE KM ST DW DD 30 10 19 14 4 15 7 H H H 'v-t NHr-suh-CD .5 4 3 7 H «3000: (ADM-503w waHH-fib N N HH MW 41 20 22 4 3 2 H LA 21 11 NHmI-‘O, 0) CW 15 16 10 7 1 8 3 (DRUIQDhi-HHQ NNHH N 0‘01me H u—u—u—s Total 62 39 96 61 100 137 63 106 68 Total 148 133 42 23 20 18 18 16 H‘h‘hihbh‘h.h‘htbib.h‘h‘hDthDhD030303090303030b¢b$ba505‘3~3‘303 138 (MOVEMENT-CHANGE IN ATTRIBUTIVE STATE Verbs KM ST DW DD MW LA CW EC Total 1 1 4 49 4 l 39 teat 1 4 17 1 make 10 15 cook 1 6 open 1 1 1 3 13 kill 12 12 break 3 shake fight I paint mess 2 bite I close 4 3 cut 6 wash 4 2 clean 6 scare 4 pep (popcorn) 3 fix 1 wrap take (a bath) 3 bust 3 drink 3 whip 3 waste 3 choke 3 turn (light on) 1 1 stir hit 1 feed put (together) polish 2 beat 2 90p (your hand with elastic) 2 tease 2 get (ready) 1 scratch 1 tear 1 cover 1 wipe 1 fill 1 smoke 1 tie 1 crack 1 fall (apart) 1 mix 1 bathe 1 knock (you out) 1 look 1 {mock (i.e., tease) I teach 1 hurt H0000.» QHHQ H O) motor-s01 ... NH N 0) his)»; H C than-AN NNHH NNNNNNNwwwwwwhD-fiwmdfimflmw HHHHHHHHHHHHHHHHHNN H Total 17 20 67 45 56 25 41 12 139 MOVEMENT-CHANGE IN POSSESSIVE STATE verbs KM ST DW DD MW LA CW EC Total tget 11 20 17 8 5 6 3 3 73 tgive 8 2 8 10 12 3 4 3 50 find 2 3 15 20 roll 9 9 buy 3 3 2 8 take 1 1 1 3 6 bring ' 5 5 catch 1 3 4 see (give it) 4 4 lose 1 1 1 3 leave 2 2 pay 2 2 pick 1 1 steal 1 1 Total 21 33 33 19 18 26 27 11 140 MOVEMENT-NONCHANGE OF STATE verbs KM ST DW DD MW LA CW EC Total look 14 15 18 25 10 15 100 tplay 30 3 14 2 17 10 3 81 tsay 10 3 4 12 2 tell 2 10 6 1 help 2 2 5 show - 1 2 1 knock (on door) call 3 1 cry cheat laugh check (heart beat) messing (fooling) rub 4 dance 3 talk 1 1 read take (a picture) 2 ask 2 holler 1 1 pee 2 sing 1 feel 1 peep taste 1 reach 1 use 1 work 1 kiss 1 2 23 58 6 5 OSNNOD OD ”‘00) 3 NOIHHQv—I .5 uu.>ra o: H H H HHHHHHHHNNNNNNwéhb-bmmmd Total 73 9 52 45 36 6O 53 25 Verbs stop get (cold) die go to sleep learn finish freeze burn mks (me sick) (school) open come (on TV) coming (badly) win boil Total Verbs wait stay lay sit hold still stand hold step hide Total 141 MOVEMENT—CHANGE OF STATE KM ST DW DD MW LA CW EC Total 2 1 3 l 4 3 14 3 1 1 5 1 4 5 3 1 4 3 3 1 1 2 2 2 3 1 4 1 1 1 1 1 1 1 l 1 l 1 1 NONP’DVEMENT-NONCHANGE POSITIONAL STATE KM ST DW DD MW LA CW EC Total 1 4 7 1 13 l 2 4 5 12 4 4 1 1 1 3 1 1 1 l 1 1 1 l 1 1 142 NONMOVEMENT-NONCHANGE EXTERNAL STATE Verbs KM ST DW DD MW LA CW EC Total watch 5 2 4 2 13 dream 1 7 8 think 1 6 7 ignore 6 6 bet 2 1 2 sleeping ~ 2 2 renmmdxur 2 2 Total 0 3 5 3 1 11 17 0 NONMOVEMENT—NONCHANGE POSSESSIVE STATE Verbs KM ST DW DD MW LA CW EC Total get (i.e., keep) 3 1 4 hold 1 l 1 3 keep 3 3 save 1 1 2 Total 1 0 7 0 2 l 1 0 APPENDIX I ACTION VERBS PRODUCED BY THE NOMAL AND LANGUAGE-IMPAIRED GROUPS APPENDIXI ACIIOIVERBSHDDIDEDBYTHEWLAMDIANCRJAGE-DIPAIREDWJPS WWI-aim IN ILXJATIVE STATE Verbs DB AD PM JK T0158]. tput 45 6 29 5 75 *go 11 5 17 11 44 fast 21 1 14 5 41 take 24 2 6 32 come 4 2 7 3 16 junp 2 5 8 15 knock 12 2 14 drive 4 1 5 10 fall 1 1 7 9 push 4 5 9 throw 4 4 8 fly 2 3 1 6 bring I 3 4 pick 1 l 1 3 drop 1 2 3 turn 3 3 walk 3 3 sit (down) 1 1 2 run 1 1 2 dunp 1 l 2 move 2 2 stand (up) 2 2 pull 2 2 swim 2 2 park 1 1 leave 1 1 stick 1 1 mp (Up) 1 1 backing up 1 1 race 1 1 land 1 1 lump 1 I tip 1 1 tramp 1 1 bans (Up) 1 1 climb 1 1 crawl 1 1 Total 142 30 103 47 143 144 'MOVEMENT—CHANGE IN ATTRIBUTIVE STATE verbs DB AD MW JK Total 00 oo eat 20 *open 3 4 tcut 5 1 fire (a gun) crash 4 paint 3 put (a hole in it) 3 clean 2 seal 2 get (dirty) 2 tickle 2 break 2 close 2 bust 1 turn (on) 1 fill 1 roll 1 slam 1 explode I spread set (a table) I hook (together) 1 fit 1 make 1 look 1 hit 1 poke 1 wake 1 KINOJIO tho H .5 H b—l HHHHHHHHHHHHHHHNNNNNNww-FQO Total 48 7 28 23 verbs tget take give pass leave Total Verbs *play say help look cry sing tell talk wave reach Total verbs go to sleep Total 145 bffimmENHPCHANGE IN POSSESSIVE STATE DB AD MW JK 17 1 10 3 1 1 2 2 2 20 1 12 6 MOVEMENT-NONCHANGE OF STATE DB AD MW JK 3 1 7 12 5 5 1 6 1 1 1 1 1 2 1 1 1 l 1 13 2 23 14 NONMONEMENT-CHANGE OF STATE DB AD MW JK 3 3 O O 0 Tbtal NNNNH Total N to 0—5 HHHHHNNQGH Total Verbs wait sit stand stay lay hold Verbs watch sleep pretend verbs keep hold save 146 NONMOVEMENT-NONCHANGE POSITIONAL STATE DB AD MW JK Total 4 4 2 1 3 3 3 3 3 1 1 2 1 1 1 1 Total 3 1 13 O NONMOVEMENT-NONCHANGE EXTERNAL STATE DB AD MW JK Tbtal 3 l 4 1 1 1 1 Total 2 0 3 1 NONMOVEMENT—NONCHANGE POSSESSIVE STATE DB AD MW JK Total 5 l 6 1 1 2 2 2 Total 7 l 0 2 LIST OF REFERENCES LIST OF REFERENCES Benedict, H. (1979). Early lexical develoment: Comprehension & production. ' Ml of Child M, 6, 183-200. Bloom, L. (1970). Me develop_nent: Found fmfition in W. Cambridge: M.I.T. Press. Bloom, L. (1975). (he word at a time. The Hague: Mouton. Bloom, L., & Lahey, M. (1978). We develog_nent;a_nd 1mg disorders. New York: John Wiley. Bloom, L., Lightbown, P., 8: Hood, L. (1975). Structure and variation in child language. Monograms of the Society for Research in Child Developnent, 4 (No. 160). Bowerman, M. (1973). Structural relationships in children’s utterances: Syntactic or semantic? In T. Moore (Edd, C_ognitive develonpentfiand the acquisition of 1mg. New York: Academic Press. Brandstadter, J. (1984) . Action developnent and developnent through action. Humn Develo t, _2_7, 115-118. Brown, R. (1973). A first We: The early stages. Cambridge: Harvard University Press. Chafe, W.L. (1970). Meangg’ and thertructure of 1mg. Chicago: The University of Chicago Press. 147 148 Chomsky, N. (1965). Aspects of the theory of syntax. Cambridge: M.I.T. Press. Clark, E.V. (1974). Some aspects of the conceptual basis for first language acquisition. In R.L. Schiefelbusch & L.L. Lloyd (Eds.), nggggge4perspectives--Acquisition, retardation and intervention (pp. 105-128). Baltimore: University Park Press. Clark, E.V. (1973). What’s in a word? 0n the child’s acquisition of semantics in his first language. In T.E. Moore (Ed.), Qggnitive development and the acquisition of lagguggg (pp. 65— 110). New York: Academic Press. Coggins, T.E. (1979). Relational meaning encoded in the two-word utterances of stage 1 Down’s Syndrome children. Journal of Speech gnd Hearing:Besearch, 22, 166-178. Duchan, F.J., & Erickson, J.G. (1976). Normal and retarded children’s understanding of semantic relations in different verbal contexts. Journal of Speech and Hearing Research, 19, 767-776. Edwards, D. (1974). Sensory motor intelligence and semantic relations in early child grammar. Coggition, 2(4), 395-434. Fokes, J., & Konefal, J. (1981). Children’s use of four semantic cases in two conditions. Journglgof Communication Digorders, 14, 497-506. Freedman, P.P., & Carpenter, R.L. (1976). Semantic relations used by normal and language-impaired children at stage 1. Journal of Speech and Hearinngesggrch, 19, 784-795. 149 Gentner, D. (1978). On relational meaning: The acquisition of verb meaning. Child Development, 48, 988-998. Goldin-Meadow, S., Seligman, M.E., & Gelmn, R. (1976). Langmge in the two-year-old. Eggnition, 4, 189-202. Huttenlocher, J. (1974). The origins of language comprehension. In R.L. Solso (Ed.), Theories in cggnitivegpgychology. New Yerk: Halsted. Huttenlocher, J., & Lui, F. (1979). The semantic organization of some simple nouns and verbs. JourrLail of Verbal mm Verbal Behaviour, 48, 141-162. Huttenlocher, J., Smiley, P., & Charney, R. (1983). Emergence of action categories in the child: Evidence from verb meanings. Psyghological Review, 99, 72-93. Johnson, J.R. (1982). Interpreting the Leiter IQ: ‘Performance profiles of young normal and language-disordered children. Journal of Speech gpd Hearigg Researc , g§, 291-296. Johnson, J.R., & weismer, S.E. (1983). Mental rotation abilities in language-disordered children. goggggl of Speech and Heggigg Research, 24, 397-403. Kamhi, A.G. (1981). NOnlinguistic symbolic and conceptual abilities of language-impaired and normlly developing children. Journal of Speechggpd Hearigg Researc , 24, 446-453. Kamhi, A.G., Catts, H.W., Kbenig, L.A., & Lewis, B.A. (1984). Hypothesis testing and nonlinguistic symbolic abilities in language-impaired children. Journal of Speech and Hearing Disorders, 49, 169-176. 150 Leiter, R.B. (1959). Part 1 of the manual for the 1948 revision of the Leiter International Performance Scale. Enychological Service Center Journnl, 11, 1-72. Leonard, L.B., Bolders, J.G., &.Miller, J.A. (1976). An examination of the semantic relations reflected in the language usage of normal and language-disordered children. Journnl of Speech and Hearing Research, 49, 371-392. Miller, G.A., & Johnson-Laird, P.N. (1976). L§Q£E§E§_§Q§ perception. Cambridge: Cambridge university Press. Morehead, D.M., & Mbrehead, A. (1974). From signal to sign: A Piagetian view of thought and language during the first two years. In R.L. Schiefelbusch & L.L. Lloyd (Eds.), LEQSEEEQ pernpectives--Acquigition, retardation, and intervention (pp. 153-190). Baltimore: University Park Press. Nelson, K. (1986). Event knowledge: Structure nnd fnnction in development. Hillsdale: Erlbaum. Nelson, K. (1985). Mnning sense: The nnguisition of shared meaning. New York: Academic Press. Nelson, K. (1973). Structure and strategy in learning to talk. Mgnngnnphs of thg Society for Rengnnnh in Child Develonngnn, 38 (No. 149). Nelson, K. (1982). The syntagmatics and.paradigmatics of conceptual development. In S.A. Kuczaj (Ed.), We develom_en_t (Vol. 2). Hillsdale: Lawrence Erlbaum Associates. 151 Nelson, K., Kamhi, A.G., & Apel, K. (1987). Cognitive strengths and weaknesses in language-impaired children: One more look. Journnl of Speechgnnd Hearing Dngorders, 52, 30-36. Orkin, M., & Drogin, R. (1974). Vitalngtatignics. New YOrk: McGraw-Hill. Piaget, J. (1952). The origins of intelligence in children. New York: International Universities Press. Rodgon, M.M., Jankowski, W., & Alenskas, L. (1976). A multi- functional approach to a single-word usage. Jounnnl of Child nnngnngg, 4, 23-43. Savich, P.A. (1984). Anticipatory imagery ability in normal and language-disabled children. Journnl of Speech nnd Henning Research, 21, 494-501. Schlesinger, I.M. (1971). Production of utterances and language acquisition. In D.I. Slobin (Ed.), The ontogenesis of gm (pp. 63-102). New Ybrk: Academic Press. Schlesinger, I.M. (1974). Relational concepts underlying language. In R.L. Schiefelbusch & L.L. Lloyd (Eds.), LEQEEEKQ perspectives -- Acouigition. retardation and intervgntion (pp. 129-152). Baltimore: university Park Press. Schwartz, R.G., & Leonard, L.B. (1984). words, objects, and actions in early lexical acquisition. Journal of Speech nnd Kenning Research, 21, 119-127. Senders, V.L. (1958). Measurement and statistics. New York: Oxford university Press. 152 Slobin, D.I. (1973). Cognitive prerequisites for the development of grammar. In C.A. Ferguson & D.I. Slobin (Eds.), Studies of child lnngnnge development. New YOrk: Holt, Rinehart, & Winston. Sldbin, D.I. (1971). The ontogenegis of gnnnnnn. New YOrk: Academic Press. Stark, R.B., & Tallal, P. (1981). Selection of children with specific language deficits. gnnnnnl of Speech:nnd Hearing Disorders, 46, 114-122. Stockman, I.J. (1986). New directions in the treatment of severe developmental disability: St. Gnllen, Switzerland’s model of guided movement thernny. Fellowship Report, Michigan State University, East Lansing. Stockman, I.J., & vaughn-Cooke, F. (1987). A closer look at children’s locative action utterances. Paper presented at the American Speech-Language—Hearing Association’s Convention. New Orleans, November. Stockman, I.J., & vaughn-Cooke, F. (1984). A closer look at the dynamic and static locative distinctions. Paper presented at the Third International Congress on Child Language. university of Texas, July. Stockman, I.J., &,vaughn-Cooke, F. (1982). Semantic categories in the language of working-class Black children. In C. Johnson & C. Thew (Eds.), Thegproceedingg of tng_necond internntional angness for the study of child langnngg, 4, 312-327. 153 wain, 0., & Yule, W. (1983). Imaginative play in language disordered children. Britigh Jou_rnal of Disorders of _Qenmmication, 1_7_, 83-92. Vygotsky, L.S. (1962). Thongnt and lmge. Cambridge: M.I.T. Press. Wren, C.T. (1982). Identifying patterns of processing disorder in six-year-old children with syntax problems. British Journal of Disorders of Congunicntjon, 1_7, 83-92. Ilu'imm' 2548 H S” H III III); II 'Imumlmj 303 69 El