37 Story Grammar Analysis ofPre-schoolers' Narratives: An Investigation into the Influence of Task Parameters Kerstin M, Tonsing and Herman Tesner Department of Communication Pathology University of Pretoria ABSTRACT This study aimed at examining the influences of the parameters of the narrative task (administered to the pre-schooler) on narrative structure. Seventeen pre-school children were selected as subjects. Five narrative·tasks were administered to each subject. The narratives were analysed for length and structure. Mainly two factors were found to influence the length and structure of the produced narratives; firstly, the presence or absence of a 'model' on which the child could base his / her narrative, and, secondly, the structure inherent in this 'model'. The implications of the study for clinical assessment ofnarra- tive skills as well as for further research are considered. OPSOMMING In hierdie studie is gepoog om die invloed van die parameters van die narratiewe taak (voorgele aan die voorskoolse kind) op die struktuur van die narratiewe produksie te ondersoek. Sewentien voorskoolse kinders is as proefpersone gekies. Vyf verskillende take is deur elke kind uitgevoer. Elkeen van die take het die vertelling van narratiewe vereis. Alle geproduseerde narratiewe is geanaliseer ten opsigte van lengte en struktuur. Veral twee faktore het 'n invloed gehad op die lengte en struktuur van die geproduseerde narratiewe; eerstens, die teenwoordigheid of afwesigheid van 'n 'model' waarop 'n kind sy / haar narratief kon baseer, en tweedens, die inherente struktuur van hierdie 'model'. Die kliniese implikasies van die bevindinge met betrekking tot die evaluasie van storiegrammatika asook die implikasies vir verdere navorsing word bespreek. KEY WORDS: story grammar, pre-schoolers, narrative task, story generation, story retelling, personal experience narra- tives INJ'RODUCTION / " Since the 1980s researchers in childhood language as- sessment have increasingly shifted their focus from the evaluation of discrete lin~istic skills on word and sentence I level to a more holistic diagnosis of longer linguistic units, such as discourse (Merritt & Liles, 1987). Amongst other types of discourse assessmknt, the evaluation of narratives has gained popularity as 'a clinical tool. Story grammars have been developed by various authors to describe narra- tive structure (Mandler & Johnson, 1977; Stein & Glenn, 1979; Roth & Spekman, 1986). The assumption underlying all of these grammars is that ideal stories are constructed of episodes, these being well-organised structures which describe goal-based behaviour. Story grammar models seek firstly to judge the completeness and well-formedness of an episode, and secondly to describe relations between episodes in multiple-episode narratives. One example 'of such a story grammar model is that described by Stein and Glenn (1979). It includes a description of story components making up an episode, as well as the relations linking these episodes to- gether. This story grammar model and various adaptations of it have been used to analyse the structure of narratives generated by subjects from the age range of pre-school to adulthood (Merritt & Liles, 1987; Peterson & McCabe, 1983). Used as a clinical tool, story grammar analysis pro- vides valuable information, not only on language abilities but also on higher level processing skills. Inferences on the cognitive organisational skills of the individual have proved to be valuable sources of information to speech-language pathologists, educators and psychologists. In a pre-school setting, story grammar analysis thus seems an invaluable clinical tool for valid and accountable language screening and evaluation, as potential learning and language-learn- ing problems can be detected at an early stage (Hemphill, Feldman, Camp, Griffin, Miranda & Wolf, 1994). The clinician intending to do such an analysis is faced with an array of choices pertaining to the variables of the task used for sampling story grammar skills. Scott (1988) lists and describes some of the parameters of narrative type and context. Amongst these are: • Source: Narratives can be self-generated or retold • Experience base: Narratives can be based on personal experiences or they can be fictional creations • Eliciting stimuli: Different types of stimuli, such as pictures or props can be made avail- able to the child. Die Suid-Afrikaanse Tydskrifvir Kommunikasieafwykings, Vol. 46, 1999 R ep ro du ce d b y Sa bi ne t G at ew ay u nd er li ce nc e gr an te d by th e P ub lis he r ( da te d 20 12 ) 38 Regarding the source of the narrative, story retelling is generally believed to be the 'easier' task, as direct input of pre-structured content is given (Liles, 1993). Literature supports the notion that story generation is a more sensi- tive indicator of story grammar abilities. Roth and Spekman (1986), for example, found significant differences between the story grammar abilities oflanguage-learning disabled subjects and subjects with normal language abil- ity as displayed during story generation tasks. Research by Peterson and McCabe (1983) suggests that narratives based on personal experiences. contain more story structures and.more mature ones than fictional nar- ratives, especially where younger children are concerned. They propose that real events need less mental ordering and may have been reported before, which leaves children to devote their attention to structural organisation of the narrative. However, children exposed to traditions of story telling have been shown to generate well-structured fic- tional narratives (Alant, Tesner & Taljaard, 1993; Kay- Raining Bird & KluppelVetter, 1994). Differences between the structural complexity of personal experience narra- tives as opposed to fictional narratives thus remain to be researched within given populations. . The types of stimuli used in eliciting fictional narra- tives in pre-schoolers differ with regard to the amount of structure inherent in the stimulus, and its potential ef- fect on the organisation of the story construction. Pup- pets, for example, do not provide the child with any struc- ture while producing a story. A single action picture pro- vides a medium amount of structure as the setting, the potential story characters and an event are given. A se- quence of related pictures gives a high degree of struc- ture. Research findings on the influence of media input on story retelling varies. In some instances visual stimuli augmenting the audio-input has been shown to aid re e call, (Liles, 1993) while others report an opposite effect (Schneider, 1996). Different effects of the use of props during presentation of the story on the one hand versus recall of the story on the other hand are also issues that have not been fully researched in the pre-school popula- tion. Insufficient and controversial data thus exists re- garding the various methodological issues involved in evaluating story grammar abilities in precschoolers - be it the source of the produced narrative (generated ver- sus retold), the experience base ofthe narrative (personal experience versus fiction) and/or the type of stimuli used for each task. This study therefore aims' to address the following questions: • Does a story grammar analysis of pre-schoolers' narra- tives elicited under various sampling conditions indicate structural similarities of productions across task? • Are there differences in productions - correlated to these. param~ters of sampling - that have practical implica- tionsin tenns of clinical asse,ssment? METHOD AIM The aim 9f this study was to investigate how pre- schoolers' use of story grammar is affected by the following parameters of the narrlltive task: the experience base, the soUrce and the eliciting.stimuli. Sub-aims included the de- scription of structural characteristics (length and story Kerstin M. Tonsing and Herman Tesner grammar parameters) of narratives elicited under five dif- ferent sampling conditions, and examining possible corre- lations and differences between them. DESIGN A cross-sectional quasi -experimental design was chosen. Children's use of story grammar was measured in five tasks. Correlations between the sampling parameters and the structural properties of the pro.duced narratives were then established. SUBJECTS A non-random convenience sample of seventeen subjects was drawn from two nursery schools. Thirteen children attended the preparatory class of an English nursery school in an affiuent area, while a further four children were non- randomly selected from the preparatory class of an Eng- lish-medium nursery school in a less affiuent area of the same town. The latter four children were intentionally cho- sen on the basis oftheir background; their teacher described them as coming from lower socio-economic backgrounds and disrupted families. Although a comparison between the two test populations, was not the aim ofthe study, a varied test population might promote a richer description of the phe- nomena (Peterson, 1994). The age range displayed by the subjects ranged from 4 years, 8 months to 6 years, 4 months, with a mean age of 5 years, 6 months. All subjects had nor- mal vision and hearing and had not received speech/lan- guage intervention before. The home language of all the children was English. Eleven boys and six girls partici pated. In order to gather infonnation on the subjects' exposure to stories and the story reading practices of their care-givers, a questionnaire was filled in by a care-giver of each of the subjects. From the results it seemed that all subjects were exposed to stories from books and from television. MATERIAL Task 1: Personal experience narrati~es . I A 'conversational map' based on McCabe and Rosenthal Rollins's guidelines (McCabe & Rosenthal Rol~ins, i994) was devised in order to elicit personal experience narra- tivesfrom the subjects. The adapted 'conversational inap' consisted of six personal anecdotes followed by variou~ sets . • I of probe questions. Each set of questions c,onsisted of one or two closed questions to establish whether a theme w'ould be successful, followed by an open question which in~ited the actual telling of the narrative. Task 2: Story generation with puppets Five cardboard puppets, as well as three cardboard pic- tures of objects were used. Task 3: Story generation with a picture / A magazine, picture was chosen as a .stimulus for this story .generation task. It depicted a setting and potential story characters involved in a problem-solving attempt. Story retelling~ 'Task 4 (Puppet Condition) and Task 5 (Picture Condition) / The South African Journal of Communication Disorders, Vol. 46, 1999 R ep ro du ce d b y Sa bi ne t G at ew ay u nd er li ce nc e gr an te d by th e P ub lis he r ( da te d 20 12 ) Story grammar analysis of pre-schoolers' narratives: An investigation into the influ~hce oitask parameters Material used for the retelling tasks consisted of two model stories as well as the visual material used for each story. In order to limit variables as much as possible, care was taken to ensure that the model stories corresponded to each other in length, structure and content. The first model story (used in Task 4) consisted of251 words and 41 clauses; while the second model story (used in Task 5) consisted of 255 words and 41 clauses. The structure of each story was analysed according to Stein and Glenn's story grammar analysis (Stein & Glenn, 1979).46 and 47 story components were identified in the two stories respectively. Both stories were classified as multiple structure episodes. A set of eight cardboard puppets as well as five c'ardboard pictures of objects were used for the first retelling task (Task 4). A se- quence of eight hand-drawn and coloured-in pictures was used for the second retelling task (Task 5). PROCEDURE A pilot study was undertaken in order to determine whether the material used during the narrative tasks was age-appropriate, and whether elicitation procedures and analytical procedures were successful. The procedures for the main study were then adapted based on the results of the pilot study. ,Pre-recording procedures: The researcher spent an average of two hours spread over three to four days per week for three consecutive weeks to establish rapport with the children. Recording procedures: Each child completed all five tasks in one sitting and was tested individually in a quiet venue at the nursery school by a single examiner. Generated and retold stories were audio-recorded by means of a Tandberg Audio Tutor TAT 771 tape recorder and TDK IEC:rtrype 1 D 60 cassettes. The production and retelling tasks were presented in this prder to eliminate the possibility of the model stories influ- ·'encing the child's generated stories. No time limits were imposed on any of the tasks. Task 1: Collection of P 1rsonal experience na~a~ives The conversational map was used. The researcher en- gag:d in conversation +ththe subject and inserted the varIOUS personal anecdotes after each of which were in- serted. two or more questions/prompts inviting subjects to share their experience. While the subje~t was narrating, merely neutral sub-prompts such as the exact repetition of a portion ofthe child's utterance, vocalisations ('ah!', 'hmm', 'oh') and neutral questions (such as 'And then?') were used. At least five narratives were collected per subject. Task 2: Story generation with puppet~ Subjects were presented with the cardboard pictures and puppets. The following instructions were given: 'Use these pictures to make up your own story. Use as many of the pictures as you like; you need not use all the pictures:Think about your story first, and tell me when you are ready.' While subjects were narrating, once again only neutral sub- prompts were given. I 39 Task 3: Story generation with an action picture Subjects were presented with ,an action pict~~e. They were given the following instructions: 'Look at this picture carefully. Make up a story about it. Your story need not be only about the picture, you can tell me about what hap- pened before and after as well. Think about your story and tell it to me when you are ready.' Tasks 4 and 5: Story retellings Subjects were given the following instructions before both retelling tasks: 'I am going to tell you a story n~w. Listen carefully. Mter I've finished, I want you to tell the story back to me.' Cardboard puppets 'and pictures of ob- jects were presented to the subject as the first model story (Puppet Condition) was narrated by 'the examiner. Follow- ing the presentation, the examiner gave the following in- structions: 'Now you tell me the story.' The cardboard pup- pets and pictures were given to the subject to use'during hislher retelling. During narration of the second model story (Picture Condition), the subject was 'presented with a se- quence of eight pictures depicting the 'story episodes. The picture~ were presented one by one 'while the examiner narrated the model story. The subject 'was thEm prompted to retell the story with the same instr'uctions as for Task 4. The pictures were not available to the subject during the retelling task.' ' " PROCESSING OF DATA , Transcription As the researcher was interested in the structural prop- erties of the children's narratives, transcription of the taped narratives focused on the content of the child's utterances. Repetitions of words and vocalisations were not transcribed. Analysis Quantitative and qualitative data w~s obtained and analysed by means of descriptive statistics - consisting of means, percentages and relative frequency measrire~"ex­ pressed as percentages (Caswell, 1995). For narratives to be included in the analysis they had to consist of ~~ini­ mum of two statements. All included narratives 'were ana- lysed for length as well as for their structural properties. Eachnarrative was segmented into statements; defined as main or subordinate clauses. Thecriteri&governing the selection of,statements for the analysis are given in Appen- dix 1. A word count was done for each ofthe elicited narra- tives. The words of all statements that were included in the analysis were counted. No fillers, repetitions or false starts were counted. A statement count was also done for each ofthe elicited narratives. Statements were then clas- sified into story components. Descriptions of these compo- nents can be found in'Appendix 2. Mter statements were , coded, so-called structural patterns (Peterson & McCabe, 1983) were identified, by which narratives were divided into sequences (non-goal-based) or episodes (goalcbased). De- scriptions ofthese structural patterns can be found in Ap- pendix 3. The data analysis consisted of comparing the narratives elicited by each of the five'tasks with respect to the following: • average length of narratives as calculated by word- and statement counts ' Die Suid-Afrikaanse Tydskrifvir Kommunikasieafwykings, Vol. 46, 1999 R ep ro du ce d b y Sa bi ne t G at ew ay u nd er li ce nc e gr an te d by th e P ub lis he r ( da te d 20 12 ) 40 • average length of structural patterns: basic structural patterns and complete episodes • frequency with which the 11 structural patterns oc- curred. Scoring reliability Eighteen (10%) of the elicited narratives were independ- ently rated by Ii second examiner, yielding a point-to point agreement of 86% for the assignment of structural patterns to the narratives. Point-to-point agreement was 85% for classifying statements into story components. RESULTS AND DISCUSSION LENGTH Analyses pertained to the average length of narratives as well as to the average length of complete episodes. Aver- age length was determined by the mean number of words, the mean number of statements and the mean number of basic structural patterns across all narratives produced for a specific task. The results are summarised in Table 1. While stories generated in response to the picture (Task 3) were similar in length to story retellings (Tasks 4 and 5), per- sonal experience narratives were signifIcantly shorter on average. Stories generated in response to puppets were sig- nificantly longer on average. The experience base of the narrative clearly influenced length, as can be established when comparing personal experience narratives (Task 1) and generated stories (Tasks 2 and 3). Personal experience narratives, by the nature of the prompts, tended to be accounts focusing on usually one dramatic incidence, with between one and three prominent characters. More than half of these narratives (55,36%) consisted of only one structural pattern. Usually no ex- tended recounts of preceding or following events was given. In contrast, the stories generated in response to puppet and picture stimuli were more elaborate narratives, includ- ing more detailed descriptions of settings, ini tiating events, attempts etc. Orsolini (1990) remarks on the more formal and elaborate style encountered in children's fantasy nar- ratives when compared to their 'real' (personal experience) narratives. He ascribes this difference to the fact that per- Kerstin M. Tonsing and Herman Tesner sonal experience narratives are aided by shared knowledge between speaker and listener, conversational co-operation and memory of the events narrated. As a result, the proc- ess of conveying pieces of verbal information is facilitated, and the speaker is allowed certain 'short-cuts' in his lan- guage use. Fictional narratives, however, require extension of discourse topic without these scaffolds, necessitating more detail and descriptions and more formal language use (Orsolini, 1990). Generated stories necessitate a great deal of planning, both of semantic content and of so-called 'macro-' and 'micro-units' (i.e. story grammar and sentence grammar respectively) (Orsolini, 1990). However, it seemed that the subjects of this study were not always aware of this demand, and spontaneously launched into all narra- tive tasks without much pre-consideration to either struc- ture or content. If the narrator was offered some form of structured 'model' (be it a highly structured picture, a per- sonal experience that can be retrieved from memory or a model story) this lack of planning could be compensated for by resorting to the structure inherent in this 'model'. However, if no structured 'model' was available (as was the case for stories generated in response to the puppet stimuli), this lack of planning resulted in long, rambling, unfocussed narratives as were produced for Task 2. When considering the two story generation tasks (Tasks 2 and 3), the degree of structure inherent in the stimulus influenced length. Puppets, being unstructured stimuli, re- sulted in elaborate, yet somewhat rambling stories. Subjects narrated consecutive actions and events, without one main goal or focus within the narrative. The single action picture, in contrast, strongly suggested a central incident for the narrative - the retrieval of a kite. Only three subjects ex- tended the content oftheit narrative beyond this incident. Regarding the retelling tasks, stories retold with the use of puppets (Task 4) were found to be longer on average than those retold without visual support material (Task 5). These results were in agreement with the research review by Liles (1993), who tentatively concludes that children respond favourably to visual support during retelling. STORY GRAMMAR ANALYSIS After coding statements into story grammar components, structural patterns were identified, and the relativ;e fre- TABLE 1: Word and statement counts of narratives produced for Tasks 1 - 5 I Task 1: Task 2: Task 3: I Task 4: Task 5: Personal Story Story Story I Story , experience generation: generation: retelling: retelling: narratives Puppet~ Picture Puppets Picture Sum of narratives produced 112 17 17 17 17 Mean number of words per narrative 58.5 232.53 95.76 111.76 90.29 Mean number of statements . 'per narrative 9.67 39.53 15.41 19.41 /,14.06 ~ Mean no. of basic structural patterns per narrative 1.69 7.42 2.41 4.12. '. 2.65 / Relative frequency of narratives consis- ting of a single basic structural pattern 55.36% 11.76% 29.41% 0.00% 11.76% / The South African Journal of Communication. Disorders, Vol. 46, 1999 R ep ro du ce d b y Sa bi ne t G at ew ay u nd er li ce nc e gr an te d by th e P ub lis he r ( da te d 20 12 ) Story grammar analysis of pre-schoolers' narratives: An investigation into the influence of task parameters quency ofthe 11 types of structural patterns (see Appendix 2 for definitions) was calculated for each of the five tasks. Hierarchies of structural patterns as a function of task were established. Data was then grouped to allow for compari- sons of the following parameters: • Relative frequency of structures with internal causal links (i.e. all structures except for descriptive and tem- poral sequences) • Relative frequency of complete episodes (i.e. all abbrevi- ated, simple, interactive and complex episodes, as well as episodes with multiple plan application) • Relative frequency of complex episodes. Tasks were compared by grouping them into hierarchies. Results are summarised in Table 2. Parameters ofthe nar- rative task clearly influenced the narrative structure as described by these structural patterns. Narratives based on personal experiences (Task 1) contained more complex episodes than the fictional narratives (Tasks 2 and 3). In this regard, the results confirmed Peterson and McCabe's (1983) suggestion that narratives about personal experi- ence may be more complex than fictional narratives as the child reports events already ordered. The demand on crea- tivity and imagination is thus less, resulting in more time being devoted to the structuring ofthe narrative (Peterson & McCabe, 1983). Another factor that might have contrib- uted to the finding that personal experience narratives were better-structured than generated stories, could be the mo- tivation underlying each task. Narrating a personal expe- rience is usually a self-initiated attempt to share an im- portant event (Heath, 1986). The narrator must structure hislher account well in order to facilitate the comprehen- sion of the listener. In narrating fictional stories, listener comprehension is, of course equally important. However, fictional stories do not emanate from the desire to share an important event - pre-schoolers in this study seemed to be less concerned with the listener's comprehension of their fictional narratives, possibly resulting in poorer structure. Furthermore, fictional narratives are only indirectly based on the causality of the real world (i.e., the relationship be- tween cause and effect), while personal experience narra- tives have a direct model which engenders causality - pro- moting a causal structure. Generated stories differed in structural complexity ac- I cording to the stimulus material used. The picture stimu- lus (Task 3) elicited more complex and complete narratives than the puppets (Task 2) which could be attributed to the I 41 greater amount of structure inherent in this stimulus (Hedberg & Stoel-Gammon, 1986). The picture portrayed goal-based behaviour, thereby strongly suggesting to the child a single-episode narrative. Puppets, on the other hand, have no inherent structure, and thus placed high demands on the narrator with regard to content. From the results'it seemed clear that the pre-schoolers were not yet skilled in structuring narratives under such conditions where de- mands on content were high (Orsolini, 1990). The influence of eliciting stimuli on the retold stories was more difficult to determine - retelling tasks differed by more than one variable, namely the type of stimulus used during presentation and availability of visual support ma- terial for retelling. Both variables or the combination could be responsible for the difference in complexity. Stories re- told for the Picture Condition (Task 5) showed greater struc- tural complexity than stories retold for the Puppet Condi- tion (Task 4). The use of the puppets during presentation of the model story (Task 4) might have given less visual structure to the story than the sequence of eight pictures (Task 5). Children's encoding of the story structure might thus have been enhanced by the picture sequence more than by the use of puppets, resulting in more complex retellings. However, it is important to note that the children remem- bered less story information and fewer episodes for Task 5, but structured their retellings better. This led to a hypoth- esis concerning the size of the memory load for each of the retelling tasks. The notion that story grammar abilities directly guide the encoding, remembrance and integrity of recall of a model story is wide-spread in the literature (Mandler & Johnson, 1977; Page & Stewart, ,1985; Montague, Maddux & Dereshiwsky; 1990). It is therefore hypothesised, that a recall task with a greater memory load necessitates a greater reliance on story grammar abilities than a task which has a lesser load on memory. The lack of visual stimuli during the activity of retelling the story for the Picture Condition (Task 5) is hypothesised to increase memory load, resulting in a smaller number of episodes being recalled. Simultaneously a greater reliance on story grammar due to the greater memory load resulted in bet- ter-structured productions. I Comparisons regarding the structural complexity of re- told versus generated narratives showed a clear tendency towards higher complexity in the retold stories. This result correlated well with research findings on narrative gen- eration and retelling of older children with and without language impairment (Merritt & Liles, 1989). It seems that a model story serves as a guide-line which facilitates the TABLE 2: Hierarchy of tasks regarding relative frequency of structures with internal causal links, complete episodes and complex episodes Relative frequency of structures Relative frequency of complete Relative frequency of complex with internal causal links episodes episodes Task 5 80.76% Task 3 66.67% Task 5 40.74% I Task 3 78.79% Task 5 62.86% Task 4 29.27% Task 4 76.47% Task 4 51.22% Task 1 17.76% Task 1 70.75% Task 1 50.66% Task 3 11.43% Task 2 I 58.76% Task 2 33.33% Task 2 6.31% Die Suid-Afrikaanse Tydskrifuir Kommunikasieafwykings, Vol. 46, 1999 R ep ro du ce d b y Sa bi ne t G at ew ay u nd er li ce nc e gr an te d by th e P ub lis he r ( da te d 20 12 ) 42 structural organisation of a child's narrative. One reason could be that a retelling task demands less imagination and creativity than a generation task, as the model story presented the retelling task provides the content ofthe story. One might argue that, while narrative retelling reduces the demands on creativity and imagination, it certainly makes demands on memory. However, literature supports the notion that remembering story details and structuring these into a narrative are closely interrelated processes (Mandler & Johnson, 1977). In the same way as structur- ing abilities were argued to aid in the encoding and remem- brance of past events, structuring abilities should aid in both the process of encoding (and therefor.e remembering) the story, as well as in the process of retelling the story. The same type of relationship does not necessarily exist between creativity (needed to generate a story) and struc- turing abilities. Whereas demands on structuring and crea- tivity therefore seem to be cumulative, competing for cog- nitive resources, demands on memory and structuring rely on the same cognitive mechanism. CONCLUSION While assessment of narrative abilities has been shown to be a valid method of sampling language and higher-level processing skills, the pragmatic nature of the task makes the narrator's performance susceptible to the influence of ,methodological variables ofthe elicitation task (Liles et aI, 1995). In this regard, researchers (such as Johnson, 1995; Schneider, 1996) alert to the need for in-depth studies of children's narrative discourse, not only across variables such as children's age, cultural, ethnic and socio-economic backgrounds but also across diff~r~nt samj)ling conditions. Within limits, this study has aimed to address the latter issue. It is thus proposed that, bearing certain methodo- logical constraints in mind, the results contribute towards a better understanding of the relationship between the narrative task parameters and the narrative produced by children of pre-school age. In particular, the results illumi- nate the influence of the methodological parameters of the experience base of the narrative, the source of the narra- tive and the eliciting stimuli used during the task on the length and structure of the product. In so-doing, some clini- cal guidelines for assessment of pre-schoolers' narrative skills can be proposed. All five story tasks elicited extended discourse, which could be analysed according to a story grammar scheme. Several factors, however, might influ- ence the clinician's choice of task in a given situation. NARRATIVE LENGTH Stories generated in response to puppets were found to be longest, while personal experience narratives were rather short on average. While length is an asset when certain linguistic parameters in the realm of semantics or syntax need to be assessed (Merritt & Liles, 1989), transcription is laborious. STRUCTURAL ORGANISATION Pre-schoolers' narrative productions benefited (with re- gard to structural organisation) from highly structured stimuli such as the action picture or the model stories. Stimuli with little inherent structure (such as the puppets) resulted in less complex and well-organised productions. Kerstin M. Tonsing and Herman Tesner This finding was related to the higher demand on creativ- ity that 'unstructured' stimuli place on the narrator (Orsolini, 1990). Pre-schoolers with a language impairment, who can be expected to have difficulties with both creativ- ity and structuring of discourse (Liles et aI, 1995), are likely to be overtaxed by tasks requiring both skills. However, creativity associated with 'unstructured' stimuli allows for observation of other variables in the children's productions, for example the theme chosen, the character details (quan- tity of characters, description of these), and the originality ofthe story line. EASE OF ELICITATION In the present study, it was found that personal experi- ence narratives were more readily told by the children. The informal, conversational method of eliciting narratives al- lowed the clinician to establish rapport with the child. How- ever, relating personal experiences to adults is not a com- mon practice in all cultural groups (Heath, 1986). Children from different cultural backgrounds might therefore not display the same spontaneity as the present subjects when prompted to share personal experiences. RELEVANCE OF THE TASK FORA SCHOOL-READI- NESS EVALUATION As narrative abilities are strong indicators of predicted school success (McCabe & Rosenthal Rollins, 1994), assess- ment of story grammar abilities should form part of a school- readiness evaluation. In selecting a narrative task for such a purpose, the clinician should ensure its predictive validity (N~uman, 1997). Th.~ ability to structure: de:conte:xtualise:d discourse is one of the prerequisites for the acquisition of literacy. While the organisational skills required for relat- ing a personal experience are reminiscent of pragmatic con- versational competence, story generation and retelling re- quire structuring of m'ore decontextualised discourse. The latter two tasks might thus be more representative of the cognitive academic language demands placed on the child in the school context. Apart from clinical implications, results add to the grow- ing body of research data on children's narrative abil'ities and can direct further research. Individual differences: and the variability of performance within one individual could be researched by comparing performance of individual chil- dren across tasks. The influence of a child's exposure to narratives on his/her story grammar abilities could be' de- termined with more extensive surveys, field observations and interviews with teachers and care-givers. A subject pool controlled for gender could illustrate the differences be- tween the story grammar abilities of boys versus girls. The validity of the present story grammar model for analysing stories told by members of non-western cultures needs to be determined as a prerequisite for accountable interven- tion. A longitudinal study could illustrate the relationship between a child's story grammar abilities and his/her con- sequent school performance. Such data is the prerequisite to including the assessment of narrative abilitie13,ih a school- readiness evaluation. The study of children's narrative ability is an exciting and expanding field of research, of interest to many profes- sionals such as educators, clinical- and cognitive psycholo- gists, anthropologists, linguists and speech-language pa- thologists. Studies such as the present one can aid in ex- The South African Journal of Communication Disorders, Vol. 46, 1999 R ep ro du ce d b y Sa bi ne t G at ew ay u nd er li ce nc e gr an te d by th e P ub lis he r ( da te d 20 12 ) Story grammar analysis of pre-schoolers' narratives: An investigation into the influence of task parameters panding the knowledge of narration and narrative devel- opment. Ultimately, the establishment of a comprehensive set of norms for narration might be achieved, although this task is a formidable one in view of the many contextual influences on narrative ability (Johnson, 1995; Schneider, 1996). However, even without a comprehensive set of norms, understanding the variables that influence a child's narra- tive production should enable clinicians to conduct valid and accountable assessments of narrative skills and should direct their efforts in intervention. REFERENCES Alant, E., Tesner, H. & Taljaard, E. (1993). Narrative performance in context: Analysis and implications within a South African context. Child Language Teaching and Therapy, 8(2), 188-204. Caswell, C. (1995). Success in statistics. John Murray Ltd.: Lon- don. Heath, S. B. (1986). Taking a cross-cultural look at narratives. Topics in Language Disorders, 7(1), 84-94. Hedberg, N. L. & Stoel-Gammon, C. (1986). Narrative analysis: clinical procedures. Topics in Language Disorders, 7(1), 58-69. Hemphill, L., Feldman, H. M., Camp, L., Griffin, T. M., Miranda, A. B. & Wolf, D. P. (1994). Developmental changes in narrative and non-narrative discourse in children with and without brain injury. Journal Of Communication Disorders, 27(2), 107-133. Johnson, C. J. (1995). Expanding norms for narration. Language, Speech and Hearing Services in Schools, 26(4), 326-343. Kay-Raining Bird, E. & Kluppel Vetter, D. (1994). Story telling in Chippewa-Cree children. Journal of Speech and Hearing Re· search, 37(6), 1354-1368. Liles, B. Z. (1993). Narrative discourse in children with language disorders and children with normal language: A critical review of the literature. Journal of Speech and Hearing Research, 36(5), 868-882. Liles, B. Z., Duffy, R. J., Merritt, D. D. & Purcell, S. L. (1995). Measurement of narrative discourse ability in children with language disorders. Journal of Speech and Hearing Research, 38(2), 415-425. Mandler, J. M. & Johnson, N. S. (1977). Remembrance of things parsed: Story structure and recall. Cognitive Psychology, 9, 111- 151. McCabe, A. & Rosenthal Rollins, P. (1994). Assessment of pre-school narrative skills. American Journal of Speech-Language Pathol- ogy, 3(1), 45-56. Merritt, D. D. & Liles, B. Z. (1987). Story grammar ability in chil- ~ dren with and without language disorder: Story generation, story retelling and story comprehension. Journal of Speech and Hearing Research, 30(4): 539-552. Merritt, D. D. & Liles, B. Z. (1989). Narrative analysis: Clinical applications of story genhation and story retelling. Journal of Speech and Hearing Dis6rders, 54(3), 438-447. Montague, M., Maddux, C. b. & Dereshiwsky, M. 1. (1990). Story grammar and comprehension and production of narrative prose by students with learning disabilities. Journal of Learning Dis- abilities, 23(3), 190-197.i Neuman, W L. (1997). Social research methods. Qualitative and quantitative approaches. Allyn & Bacon: Boston .. Orsolini, M. (1990). Episodic structure in children's fantasy nar- ratives: "Breakthrough" to decontextualised discourse. Lan- guage and Cognitive Processes, 5(1), 53-79. Owens, R. E. (1995). Language disorders: A functional approach to assessment and intervention. Allyn & Bacon: Massachusetts. Page, J. L. & Stewart, S. R. (1985). Story grammar skills in school- age children. Topics in Language Disorders, 5(2), 16-30. Peterson, C. (1994). Narrative skills and social class. Canadian Journal of Education, 19(3), 251-269. Peterson, C. & McCabe, A. (1983). Developmental psycholinguistics. Three ways of looking at a child's narrative. Plenum Press: New York. Roth, F. P. & Spekman, N. J. (1986). Narrative discourse: Sponta- neously generated stories of learning-disabled and normally achieving students. Journal of Speech and Hearing Disorders, 51(1),8-23. Schneider, P. (1996). Effects of pictures versus orally presented stories on story retellings by children with language impair- ment. American Journal of Speech-Language Pathology, 5(1), 43 86-95. Scott, C. M. (1988). A perspective on the evaluation of school chil- dren's narratives. Language, Speech and Hearing Services in Schools, 19(1),67-82. Stein, N. & Glenn, C. (1979). An analysis of story comprehension in elementary school children. In R. O. Freedle, Editor, New directions in discourse processing, Ablex: New Jersey. APPENDIX 1 Scoring procedure: Story components The following procedural guidelines were adapted from Peterson & McCabe's (1983) story components, which was based on Stein & Glenn's story grammar categories (Stein & Glenn, 1979). If a statement met the criteria for more than one component, it was scored as both. A statement was generally not scored as more' than two categories, and maximally in three, which occurred seldom. Assigning com- ponents to statements proceeded in the same manner for both generated and retold narratives. Statements that were retold from the model story were thus assigned a certain component based on the function they served within the retold story, regardless of the component which they had been assigned in the original model story. APPENDIX 2 Story grammar components Set 1: Components of this set were assigned to statements within structural patterns displaying internal causal links. 1. Setting: Introductions of characters, descriptions of lo- cation or time, information on the habitual social, physi- cal, and!or temporal context, as well as habitual actions or states. 2. Initiating event: Any natural occurrences, actions, in- ternal events and awareness of these (e.g. pain, hunger, etc.) as well as verbalisations having some immediate result or evoking a response, internal response or at- tempt from a character. 3. Internal response: Internal states such as affects, cognitions, or goals, that motivated a protagonist and! or were caused by an initiating event. 4. Response: All verbal responses of the characters to an initiating event, provided that such responses were not an expression of internal affect, cognition or goals, in which case they fell into the category of internal re- sponses. 5. Plan: Statements specifying the character's strategy for obtaining a goal. 6. Attempt: The character's overt action(s) to obtain a goal. 7. Direct consequence: Descriptions of the character's suc- cess or failure at attaining hislher goal(s), any existing states once all attempts had failed, or any change in the sequence of events resulting from the character's at- tempts. 8. Reaction: How the character felt about the outcome, and verbal responses to this. 9. Result: A result was scored when an initiating event and! or an internal response caused a change in the environ- ment which had not been planned or intended by the character(s), and it did not result from a goal-based ac- tion from the character(s). Die Suid-Afrikaanse Tydskrif vir Kommunikasieafwykings, Vol. 46, 1999 R ep ro du ce d b y Sa bi ne t G at ew ay u nd er li ce nc e gr an te d by th e P ub lis he r ( da te d 20 12 ) 44 Set 2: Components of this set were assigned statements within narratives without causal links, or togeneral statements (which neither served a specific function nor provided set- ting information) occurring within a structural pattern with internal causal links. 1. Actions: These described the behaviour ofthe character. Included in this category were verbalisations of the character(s), such as 'He said, "Happy birthday, Lisa!'" 2. External states: This category could be described as roughly an equivalent of the setting-category, as it in- chided all statements which described the story envi- ronment, such as introductions.and descriptions of the character(s), their physical chaFaCteristics and states (hunger, pain, tiredness) and information about the lo- cation and the temporal aspects. The character's posi- tion in space was also incluci~d in this category, for ex- . amJlle, .~He stood next to her'.' \. 3. Internal states: These described the internal states of the cha~aCter(s), that is, affect, likes, dislikes and char- acter traits. 4. Occurrences: These described changes in the environ- ment and/or character which neither were character-in- duced nor fell into the category of internal states. Ex- amples are: 'The car crashed' or 'He died'. APPENDIX 3 Structural patterns in order of increasing complex- ity , 1. Descriptive sequence: Descriptions of charaCter(s), sur- roundings, and habitual actions without any temporal or causal relationship. ,Kerstin M. Tonsing and Herman Tesner 2. Temporal sequence (t / s): At least two actions arid/or oc- currences that were chronologically rather than caus- ally ordered. 3. 'Reactive sequence (r / s): In these sequences, events, changes or conditions automatically caused' other changes, but no planning or goal-based behaviour from character(s) was evident. 4. Abbreviated episode (a/e): A goal could be understood though it was inferred rather than explicitly stated. 5. Simple episode:An entire goal-based behaviour sequence was described, including a motive for action in the form of an initiating event or an internal response, an overt attempt and a direct consequence. 6. Multiple plan application: An elaboration ofthe simple episode where the character met with failure ofhislher attempt to achieve a goal, and consequently engaged in at least one new attempt to reach the same goal. 7. Interactive episode: Two characters had (usually oppos- ing) goals and influenced each other. When several structures had causal links between them, they formed a complex episode. Three types of complex epi- sodes exist: 1. Complex episode: embedded reactive sequence: A reac~ tive sequence was causally related to an episode 2. Complex episode: embedded episode: Two episodes were causally related 3. Multiple structure episode: More than two episodes and/ or reactive sequences were causally related Incomplete episode (/ e) These episodes lacked a direct consequence. If no conse- quence could be inferred, yet the character's behaviour was purposive, episodes were scored as incomplete. The South African Journal of Communication Disorders, Vol. 46, 1999 R ep ro du ce d b y Sa bi ne t G at ew ay u nd er li ce nc e gr an te d by th e P ub lis he r ( da te d 20 12 )