UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Pronominal anaphoric reference in the narratives of 3-year-old children Gomme, Norma Jean 1994

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_1994-0424.pdf [ 2.56MB ]
JSON: 831-1.0087474.json
JSON-LD: 831-1.0087474-ld.json
RDF/XML (Pretty): 831-1.0087474-rdf.xml
RDF/JSON: 831-1.0087474-rdf.json
Turtle: 831-1.0087474-turtle.txt
N-Triples: 831-1.0087474-rdf-ntriples.txt
Original Record: 831-1.0087474-source.json
Full Text

Full Text

PRONOMINAL ANAPHORIC REFERENCEiN THE NARRATIVES OF 3-YEAR-OLD CHILDRENByNORMA JEAN GOMMEB.Sc., The University of Victoria, 1991A THESIS SUBMITTED IN PARTIAL FULFILMENT OFTHE REQUIREMENTS FOR THE DEGREE OFMASTER OF SCIENCEinTHE FACULTY OF GRADUATE STUDIESTHE FACULTY OF MEDICINETHE SCHOOL OF AUDIOLOGY AND SPEECH SCIENCESWe accept this thesis as conformingto the required standardTHE UNIVERSITY OF BRITISH COLUMBIASeptember, 1994©Norma Jean Gomme, 1994In presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agreethat permission for extensivecopying of this thesis for scholarly purposes may begranted by the head of mydepartment or by his or her representatives. it isunderstood that copying orpublication of this thesis for financial gain shall notbe allowed without my writtenpermission.(Signature)Department of S’h -enzsThe University of British ColumbiaVancouver, CanadaDate &DE-6 (2/88)ABSTRACTThe general purpose of the present study was to investigate the way in which 3-year-oldchildren use pronouns to refer to characters in a narrative. A form/function approachwas taken, exploring not only the forms used to refer to story protagonists, but also thefunctions those forms served to create coherence in the narrative as a whole. The studyhad three specific goals. The first of these was to compare the results of this study withthose of Bamberg (1987), specifically to determine whether the 3-year-old English-speaking children in this study were using the same reference strategy as the 3-year-oldGerman-speaking children in Bamberg’s study (the thematic advancement strategy). Thesecond goal was to compare the children’s referential use of pronouns at times whenthey were unfamiliar and familiar with the story. The third goal was to compare theresults of this study with those of Karmiloff-Smith (1981) in an attempt to resolvedifferences between her study and Bamberg’s regarding the age at which children movefrom deictic to referential use of pronouns, and the exact nature of the thematicadvancement strategy.Ten English-speaking children between the ages of 3;2.3 and 3;9.12 participated in thisstudy. Each child told two stories, first when they were unfamiliar with the book (T1),and again after reading the book with their caregiver(s) over the course of a week (T2).Results showed that these children as a group demonstrated the developing ability tocreate textual coherence through pronominal reference in a manner consistent withBamberg’s thematic advancement strategy. Closer analysis of individual stories pointed11to the presence of several substrategies, and showed developmental variation in thechildren’s abilities to use pronouns referentially. No significant difference was found inthe children’s referential use of pronouns between T1 and T2, although otherdevelopmental measures showed improvement with story familiarity. Further, theresults did not provide evidence for purely deictic use of pronouns, or the inability tocreate any level of textual coherence as found by Karmiloff-Smith for English-speaking4-year-olds. Results also disputed Karmiloff-Smith’s proposal that the thematic subjectstrategy involves exclusive preference for clause-initial position to be reserved forreference to the protagonist. It is proposed here that differences between the studies canbe attributed to variation in experimental design.111TABLE OF CONTENTSABSTRACTLIST OF TABLES viiACKNOWLEDGEMENTS xCHAPTERINTRODUCTION AND REVIEW OF THE LITERATUREIntroductionOutline of the ThesisBackground of the StudyNarrative ProductionThe Effect of Task and Context on Chidren’sStory ProductionCohesion and CoherencePronominal AnaphoraThe Development of Pronominal AnaphoraHypotheses2. METHOD 22OverviewSubjectsProcedures for Data CollectionPart I (Unfamiliar Condition)Part II (Familiarization)Part III (Familiar Condition)ivTranscription and CodingForm CategoriesFunction CategoriesSubcategoriesMention of Other CharactersMention of Characters Accompanied by Finger PointStory Structure ElementsIndicators of Picture Description or Story LineWords Per ClauseAnalysis3. RESULTS AND DISCUSSION 45OverviewGroup DataComparison with Bamberg’s Data and ResultsDistribution of Reference Between the Two Main CharactersDistribution of Nominal and Pronominal FormsSwitch ReferenceMaintain ReferenceReference to the Boy Contrasted With Reference to the DogSummaryComparison of Performance at T1 and T.,Measures to Assess Form/Function Pairings at T1 vs. T2Nonreferential Measures of Narrative CompetenceComparison with Karmiloff-Smith (1981)The Use of Gesture (Point) with PronounsDistribution of Referents to Pronouns in Initial PositionAcross-Subject MeasuresErrors and Self-CorrectionsIndividual DataUse of Nominals to Switch Reference to the BoyUse of Pronominals to Switch Reference to the DogV4. SUMMARY AND CONCLUSIONS 102General SummarySummary of ResultsComparison with Bamberg’s ResultsComparison Between T1 and T,Joint ReferenceComparison with Karmiloff-Smith (1981)Individual DataAcross Subject MeasuresOverall Summary of ResultsExplanation of Differences Between StudiesThe Scope of the Present StudyFuture ResearchConcluding RemarksBIBLIOGRAPHY 119APPENDIXA. Frog, Where Are You? (Mayer, 1969): 122Picture by Picture DescriptionB. Sample Transcript 124C. Subcategorized Form/Function Coding Pairs 136viLIST OF TABLESTable Page1. Gender and age of subjects at time of firststory elicitation (T1) 242a. Number (and percentage) of nominal, pronominaland elliptical reference to the boy and to thedog, at T1 and T2 (joint reference to boy anddog included) 462b. Number (and percentage) of nominal, pronominaland elliptical reference to the boy and to thedog, at T1 and T2 (joint reference to boy anddog excluded) 473a. Proportions of reference in nominal form madeto each main character (boy and dog) forspecified functions at each time (joint referenceto boy and dog included) 513b. Proportions of reference in nominal form madeto each main character (boy and dog) forspecified functions at each time (joint referenceto boy and dog excluded) 524a. Comparison of proportion of pronominal referenceout of total reference made for a specified functionand time, reference to boy vs. reference to dog(joint reference to boy and dog included) 564b. Comparison of proportion of pronominal referenceout of total reference made for a specified functionand time, reference to boy vs. reference to dog(joint reference to boy and dog excluded) 575a. Comparison of pronominal reference to the boy andto the dog, for specified reference functions,T1 vs. T2 (joint reference to boy and dog included) 615b. Comparison of pronominal reference to the boy andto the dog, for specified reference functions,T1 vs. T2 (joint reference to boy and dog excluded) 62vii6. Total number of clauses and number of thematicadvancement clauses at each telling, individualand group; percentage of total clauses contributingto advancement of the story theme, individual andgroup, at T1 and T2 647. Comparison of the proportion of clauses whichcontribute to thematic advancement out of the totalnumber of clauses summed for all subjects, T1 vs. T2 648. Mean Length of Clause* for all subjects at T1 and T2 659. “Goodness” of stories, as determined bypresence/absence of six basic story elements(based on Berman, 1986), at T1 and T2 67lOa. Gestural data summed for all videotaped subjects:Number of pronominal and nominal full referentialforms and deictic terms with point, number ofdeictic terms without point, and number ofoccurrences of point with terms other than deicticor full referential forms, at T1 and T2 72lOb. Gestural data summed for all videotaped subjects:Number of pronominal and nominal full referentialforms and deictic terms with point, number ofdeictic terms without point, and number ofoccurrences of point with terms other than deicticor full referential forms, at T1 and T2 (excludingSubject 3) 7311. Comparison between summed number of deictic termsaccompanied by point and number of deictic termsunaccompanied by point, at T1 and T2, with andwithout Subject 3 7412. Utterance-initial pronouns used by each subject atT2 to refer to the boy and jointly to the boy anddog, shown as total and by switch reference, maintainreference and maintain reference (local) functions 7913. Test for positive linear correlation between acrosssubject measures: Mean Length of Clause (MLC),Goodness of Story, Age (months) 83viii14. Comparison of proportion of pronominal referenceout of total reference made to the boy serving theswitch reference function between the oldest foursubjects and the youngest four subjects, summed forT1 and T2 (joint reference to boy and dog included) 84ixACKNOWLEDGEMENTSThe completion of this thesis marks the end of a long, meandering process. At eachstage in its creation, different people have offered the kindness and support necessaryto “keep going”. I thank the children of this study, who brought spirit and innocenceto their story telling, qualities lost in the static pages of the transcripts. I thank myhusband, Graham, for his confidence and optimism, and my parents for their technical,emotional and culinary support. My deepest thanks are reserved for Dr. CarolynJohnson, whose generosity has known no limits.I dedicate this thesis to my daughter, Hannah, who has enriched my life beyond allexpectation, and whom I thank in advance for the stories she will tell when she turnsthree.xCHAPTER ONEINTRODUCTION AND REVIEW OF THE LITERATUREIntroductionThe purpose of this investigation was to examine the manner in which 3-year-oldchildren use pronouns to refer to characters within a narrative. In particular, thedifference between how children used pronouns to refer to the protagonist of a story asopposed to another main character was observed. Investigation of this specific aspectof the reference system may aid in the understanding of the functional ability of 3-year-old children to maintain cohesive structure throughout a story in order to createcoherence for the text as a whole. The young age of the children being investigated inthis study may result in evidence of a developing form/function relationship betweenpronouns and their use to serve a discourse function.This study was designed to meet three main goals. First, the study closely replicatedpart of a study conducted by Bamberg (1987), which examined this same use of referencein the stories of German-speaking children from the ages of 3;6 to 1O;1. It was intendedthat this study and Bamberg’s be close enough in design and analysis to allow directcomparison of results. From his collected data, Bamberg proposed a series ofdevelopmental stages through which children move to reach the aduitlike applications1of pronominal anaphoric reference within narratives. Therefore, the results from thisstudy will be compared with, and if appropriate, inserted into Bamberg’s developmentalscheme.Second, the data was compared with that of Karmiloff-Smith (1981), another prominentstudy into the development of pronominal anaphoric reference. Karmiloff-Smith’s studysupported a notion similar to Bamberg’s of the first stages in the development ofadultlike pronominal anaphora (with minor variation), but differed considerably withrespect to the age at which the first stage of referential use of pronouns appears.Because Karmiloff-Smith’s study was conducted with English-speaking (and French-speaking) children, it was important here to compare the two sets of English data. Anydifferences and/or similarities between the data are discussed with respect toexperimental design and analysis. The discussion is enriched by results found by otherresearchers.A third goal of the study was to observe the effect of story familiarity on the use ofpronouns to make reference. Therefore the difference in use of pronominal referencebetween a condition where children had no previous adult model of a story and acondition in which they had repeatedly received an adult model was examined. TheBamberg study focused only on the condition in which the child had received an adultmodel.2Outline of the ThesisThe remainder of chapter 1 is a review of the research literature that motivated thepresent study and a statement of my hypotheses. Chapter 2 presents the method usedto test these hypotheses. Chapter 3 presents the results of the study and a discussionof these results in the context of the research findings of other authors. Chapter 4 offersa summary of the results and conclusions drawn from them. Chapter 4 also presentsrecommendations for further study.Background of the StudyWhen we ask a child to “tell us a story,” what exactly are we asking the child to do?What knowledge and resources are we expecting the child to have in place in order toaccomplish the task? First, we expect that the child has knowledge of what a story is.In other words, we expect that at least an immature form of an adultlike story schema(as discussed by such authors as Stein & Glenn, 1982) is already part of the child’sknowledge base. We also expect the child to draw heavily on not only experience withthe basic structure of stories, but also on personal and social experience with problem-solving behaviour (including such elements as goal-setting and achievement, and thetypes of problems that occur). Further, we expect the child to be able to utilize herdeveloping language skills to produce a string of utterances that, when considered as awhole text, contain enough elements of a “good” story and enough textual organization(coherence) to be considered a story at all. Finally, we expect the child to monitor allof the above with respect to the effect the story is having on the listener. Therefore, we3are asking the child to integrate simultaneously developing abilities in the areas ofcognition, communicative competence and language in order to “simply” tell us a story.A significant body of research in the area of narrative acquisition has focused on thedevelopment of story schema knowledge in children (e.g. Stein & Glenn, 1982; Hudson& Shapiro, 1991). Through various studies, mainly examining comprehension and recall,the bulk of the data indicates that children begin to demonstrate knowledge of anaduitlike story schema between the ages of three to four years. Most research indicatesthat the development of a full story schema is not completed by most children until theyare well into their elementary school years.Less research has focused on the production of narratives and the developing ability touse linguistic elements as cohesive devices in order to produce a coherent narrative text.Narratives provide uninterrupted, child-generated strings of utterances which canprovide insight into the processes by which children choose form “the myriad ofstructures available to create coherent narrative text” (Liles, 1993: 870). Further, becausethese strings of utterances are placed in the context of a narrative genre, factors presentin the narrative context which influence a child’s choice of form to serve a givencohesive function may also be uncovered (e.g. tense and aspect, or the advancement ofa story line). Spontaneously generated narratives also produce errors and selfcorrections, which can give further information regarding the developing processesunderlying the production of coherent text.4In the following review of the literature I will first discuss some issues surrounding theconflicting experimental results in studies of narrative development. Following this Iwill briefly define the terms cohesion and coherence as they are used this study. FinallyI will discuss pronominal anaphora and its development, focusing on two pivotal studiesin this area: Bamberg (1987) and Karmiloff-Smith (1983).Narrative ProductionIn the work that has been done with narrative production, there have been conflictingresults with respect to the age at which children demonstrate knowledge of a storyschema in their productions and the age at which they can ultimately tell “coherent,goal-based, fictional stories” (Hudson & Shapiro, 1991: 102) similar to those told byadults, although the research generally places the lafter achievement well into theelementary school years. Further, there is disagreement in the literature as to the agesat which children begin to use various linguistic forms to serve specific cohesivefunctions, resulting in a connected, coherent narrative. Discussion of these differencesin results between studies which investigate similar aspects of narrative structure andorganization, or the linguistic devices used to create text coherence, has focused onvariation in task and context among studies.The Effect of Task and Context on Children’s Story ProductionFirst, the issues of narrative length and narrative complexity must be considered. If anarrative is too short, the incidence of a particular feature of the narrative may be toolow to provide useful information. For instance, if one is interested in determining the5statistical significance of the frequency of use of one form as opposed to another to filla cohesive function within the narrative (as in the present study), then low incidence ofthe forms could jeopardize the statistical significance of results and magnify theimportance of individual variation (Liles, 1993), A longer narrative may increase theincidence of particular forms and eliminate restrictions on statistical analysis. Withrespect to narrative complexity, research has indicated that the complexity of linguisticstructure within a narrative increases with an increased number of characters andepisodes (e.g. Peterson & McCabe, 1983). It follows that if a narrative is too simple achild may not be required to “actively process [the] organizational strategies underinvestigation” (Liles, 1993: 878). However, if a narrative is too complex, including notonly its number of characters and events, but also the required level of vocabulary andorganization, the resulting production may be too minimal to warrant analysis (Ripich& Griffin, 1988, as cited in Liles, 1993).Further to narrative length and complexity, the context of elicitation of a narrative is alsocrucial to the resulting production, in particular the coherence of the text. Research hasshown that if a child is familiar with the content and topic of a narrative, moredevelopmentally sophisticated production results (Hudson & Shapiro, 1991). Hudson& Shapiro conclude that “children may display more advanced cohesive devices. ..innarrative genres1 they have already mastered” because the difficulties children1 Hudson & Shapiro distinguish among fictional narratives, personalnarratives and scripts, and discuss the formal characteristics and taskdemands of each of these genres.6experience in creating coherence may detract form their abilities to use cohesive devices(1991: 106).Following from the discussion above, one must take a critical view of the experimentalmethods and resulting narratives from which data and conclusions are drawn whencomparing studies into the development of narrative structure and the linguistic formsthrough which local cohesion and textual coherence are created. With respect to thefocus of the present study, these cautions certainly apply to the investigation into thedevelopment of pronominal anaphoric reference, where similar variations in method anddata driven conclusions also exist.Cohesion and CoherenceBefore entering a discussion on the development of pronominal anaphoric reference, itis necessary to define briefly the terms “cohesion” and “coherence,” due to the widerange of interpretations these terms have entertained in the literature. The view to betaken here will reflect those of Bamberg (1987) and Karmiloff-Smith (1981). Coherencecan be defined as “that quality which makes the discourse stand as a whole text; that is,which makes the utterances of the text related to one another in some salient way, andnot just random strings” (Bennett-Kastor, 1983: 136). This coherence of a whole text isin part the result of sequential cohesion: the use of linguistic forms (cohesive devices)to create these connections between utterances. Cohesion has often been studied insentence pairs, concentrating on the function of a cohesive device in a local context.7However, local cohesion in a narrative works to produce textual coherence, and thereforethe function of such a cohesive device is expanded, if not altered.In order to recognize the influence of the narrative context on the use and function ofcohesive devices, Bamberg (1987) defines cohesion as involving the relationship betweentwo different levels: the linguistic level (at which linguistic elements cue intersententialand interclausal relationships) and the conceptual level (at which events combine toproduce units which constitute the narrative “whole”)2. According to Bamberg, a“globally coherent” interpretation is the result of both top-down and bottom-upstrategies: top-down strategies involving personal experience with the world and withnarratives that allow the listener to make predictions, and bottom-up strategies involvingthe cohesive devices that cue the context-dependent interpretation of sequentialutterances. This functionalist perspective is also reflected in the theoretical stanceKarmiloff-Smith brings to her research. -The view of coherence taken here is different from the view of many researchers in thatit is focused on the perspective of the speaker and the specific cues a speaker inserts intoa text to create coherence for a listener. Naturally, this view is tied to the speaker’sproduction of discourse and the specific linguistic forms which serve as cues. Otherauthors, such as Brown & Yule (1983) have considered coherence from the perspectiveof the listener, and the expectations of textual connectedness a listener brings to a2 In this way, Bamberg diverges from those who reserve “cohesion” to referto formal linguistic marking of semantic relations. See Halliday & Hasan(1976) and Brown & Yule (1983).8discourse context (such as a narrative). This view of coherence has been part of thetheoretical basis for many investigations into discourse abilities. It follows that thedesign of such studies is based on comprehension of discourse, not production. (SeeBrown & Yule, 1983, for further discussion. See Halliday & Hasan, 1976, for a detailedexplication of cohesion from the production point of view that is related, but notidentical, to the framework used in this thesis.)If cohesion is to be viewed from the perspective of the speaker, then it is necessary todescribe not only the linguistic forms which serve as cohesive devices, but also thefunction(s) they serve in the particular linguistic and textual context in which they occur.This is easily illustrated by the fact that many linguistic forms have plural functions,and, therefore, disambiguating the form/function relationship at a specific instance ishighly dependent on context (Bamberg, 1987). In addition to its ability to disambiguatefunction, a form/function approach to studying cohesive devices is a powerful tool forassessing the changing use of linguistic forms to serve particular cohesive functions overdevelopmental time.Pronominal AnaphoraPronominal anaphora can be generally defined as the use of pronouns as contextual cuesto indicate that sentences or groups of sentences are related and are to be jointlyinterpreted with respect to referent (Bamberg, 1987). Pronominal anaphora and itsdevelopment have been the subject of discussion in the work of various authors.Halliday & Hasan (1976) formally stipulated that pronouns on their own have no9inherent referential meaning, but rely on nominal expressions found elsewhere in thetext to gain referential meaning. However, this notion of strict reliance of pronouns onsubstitution by nominals to gain meaning has been rejected, at least in part, by severalresearchers.Brown & Yule (1983) are two authors whose interpretation of the meaning relationshipbetween a referential pronoun and its referents has moved away from Halliday &Hasan’s “substitution” view. Brown & Yule argue that a simple “substitutionrelationship” can be misleading and falls short of explaining the function of pronounsas anaphors. They suggest that the interpretation of pronominal reference in discourserelies not only on the antecedent nominal expression, but also on other linguistic andpragmatic factors. For example, in the sentence “Even an apprentice can make overtwenty pound a week and they don’t get much tax [taken] from that” (1983: 217), areference to the antecedent nominal “an apprentice” is made by the plural pronoun“they.” This would seem to be a problem of number agreement from a strict substitutionview of pronominal reference. However, when the antecedent nominal is considered inthe context of the attached predicate “can make over twenty pound a week” theantecedent nominal is interpreted as being one of a set of individuals; the followingpronoun therefore refers to the whole set.Wiese also argues that pronouns carry meaning and emphasizes that this meaning“reflect(s) social... classifications of possible referents” (1983: 373). He also emphasizesthat the referents of pronouns need not be found directly in the preceding text, but that10referents need only be assumed by the speaker to be inferred by the listener due to theirshared knowledge of the “universe of discourse” (1983: 373). In keeping with this view,Martin (working within the framework of Halliday’s systemic grammar) proposed anetwork of reference (of which pronominal anaphora is a part) which describes “theoperations a listener might perform in locating the referent of a phoric nominal group”(1985: 271). This system is dynamic and, like Wiese’s, recognizes the interplay of culture,context and text in the interpretation of reference.This move away from viewing pronominal anaphora as a strict anaphor-antecedentsubstitution relationship may be the result of researchers analysing anaphora in whole,connected texts rather than simply in sentence pairs. This further begs the analysis ofanaphora by a form/function approach, as has been adopted by such authors asKarmiloff-Smith (1981) and Bamberg (1987), who have examined the development ofpronominal anaphora in the narrative context. The narrative context is rich withanaphoric reference, due to the repeated actions and interactions of characters. From theperspective of studying pronominal anaphora it is ideal, because “not just the repetitionof an NP, but the form of its introduction and subsequent mention, and the grammaticaland semantic characteristics of it, reveal what the story is the child has in mind, as wellas his grasp of the linguistic tools available for expression of the specific discourse”(Bennett-Kastor, 1983: 147).11The Development of Pronominal AnaphoraAs was discussed above for the study of narratives in general, in the examination ofpronoun function in narrative text one of the more contentious issues has been thedevelopmental point at which pronouns stop being used only deictically and start beingused referentially. The deictic use of pronouns involves the creation of a joint,nonlinguistic focus between the speaker and listener in the “here and now,” oftenaccompanied by a pointing gesture in appropriate extralinguistic contexts (e.g. pointingto a character on the page of a picture book). In contrast, referential use of pronouns isnot linked to the “here and now,” but rather connects related propositions linguisticallyand contributes to the overall textual structure of connected discourse (Hickman, 1983;Wales, 1986). There is even discussion as to whether it is possible to separate referentialand deictic function of pronouns after referential use is acquired; the functions may beat work simultaneously (Schiffrin, 1990). However, it is not disputed that indevelopment children move from deictic to referential use of pronouns, in particularthird person pronouns, as they undergo the process of acquiring aduitlike anaphoricpronominal reference. Indeed, this change from unifunctional to bifunctional use of thepronoun in the narrative context fits with the generally accepted developmental trendin which linguistic forms are first used unifunctionally and only later move toplurifunctional use. Rather, it is the developmental age at which referential use firstappears, and the stages through which children progress toward the adult system, whichhave received differing interpretations. Although other authors have addressed thisissue, this study will focus on the research of Bamberg (1987) and Karmiloff-Smith(1981). These studies are regarded as “some of the best research into the textual context12of children’s narratives” (Liles, 1993: 873). Further, their view of the developmentalprocess toward adultlike pronominal anaphora is similar, excepting a wide discrepancyin their opinions concerning the age at which the first signs of referential use ofpronouns appear.In her study of 350 English and French children, ages four to nine years, Karmiloff-Smith(1981) defines stages of acquisition of pronominal anaphora in narrative productions.The design of the study involved showing the children books containing six pictureseach, and for each sequence of pictures asking the children to say “what is happening.”Each collection of pictures introduced a different number of characters, and thesecharacters differed in their involvement or “importance” in the story. For example, inone story type there existed one main character, with another character who wasinvolved only fleetingly (story type A). In another story type (story type C) twocharacters were present throughout. In analysing the various stories told by thechildren, Karmioff-Smith concluded that the children under the age of six were usingpersonal pronouns deictically, They seemed to be treating the pictures as separateentities, not creating a coherent text from the sequential pictures and correspondingevents. The fact that pronouns were generally accompanied by a point to the characteron the page, thereby drawing on shared nonlinguistic contextual knowledge of both thechild and the experimenter, was interpreted by Karmiloff-Smith to indicate the deicticnature of the pronoun use.13In contrast, by the time the children reached about six years of age, they began toevidence the use of pronouns in what Karmiloff-Smith labelled a “thematic subject”strategy. For the younger children (i.e. starting at six years), when this strategy wasemployed, pronouns appeared only in utterance-initial position and were reserved forreference to the thematic subject, or main protagonist, of the story. Other characterswere referred to by full nominal expressions. In this way, the child was interpreted tobe paying attention to the relationship between the depicted events, using pronouns toadvance the thematic progress of the narrative. As the children got older, they beganto locally refer to characters other than the main protagonist by utterance-initialpronouns, but they added a postposed nominal phrase to clarify the referent for thatpronoun (French data). Further, even when children employed this latter strategy, if apersonal pronoun occurred in utterance-initial position without a clarified referent, it wasalways in reference to the main protagonist, even if the last mention of that characterwas sequentially removed in the text (Karmiloff-Smith, 1981; Bamberg, 1987).This “thematic subject” strategy of pronominal reference is not a strategy employed byadult speakers. Adult pronominal reference is truly anaphoric in nature, using pronounsto follow and refer to local mention of characters regardless of their “prominence” asprotagonists. Adults use pronouns anaphorically to maintain reference to characters,using full nominal expressions to switch reference in a text as characters are alternatelyintroduced or mentioned (Bamberg, 1987). According to the Karmiloff-Smith data, thisaduitlike system of pronominal anaphora is not in place until after nine years of age.14This relatively late age for the acquisition of adultlike use of pronominal anaphora wasconfirmed in a study of similar design by Bamberg (1987). In this study of narrativesproduced by German children between the ages of 3;6 and 1O;1, Bamberg also found thatonly seven of nine children in the age range of nine to ten years utilized an aduitlikesystem of pronoun anaphora, (i.e. consistently using pronouns to serve a reference-maintaining function and nominal phrases to serve a reference-switching function).However, other results of Bamberg’s study contrast with those found in the KarmiloffSmith (1981) study, particularly with respect to the age at which the “thematic subject”strategy is first evidenced, and the details of the progression through which childrenmove as they acquire an aduitlike system.Bamberg’s study differed from Karmiloff-Smith’s in several ways, which may havecontributed to the different results and interpretations derived from the data. First, thepicture book used to elicit the narratives (Mercer Mayer’s Frog, Where Are You?) was fourtimes as long (twenty-four pictures), and therefore allowed for longer pieces of text tobe analyzed for each telling. Second, again in part due to the length of the picture-book,events throughout the book involved one or both of the two protagonists (the boy andhis dog), as well as several minor characters. Therefore, there were many chances withina single story for pronominal reference to be made with respect to multiple characters.Third, Bamberg closely analyzed the stories of individual children, recording theiridiosyncratic uses of pronouns as well as their uses in common with other children. Indoing so, Bamberg was able to more finely describe the stages in the acquisition ofpronominal anaphora that the children in his study appeared to exhibit.15By examining the linguistic devices through which the children maintained and switchedreference with respect to the two main characters of the story, Bamberg found adevelopmental trend as follows. The younger children almost exclusively chose toswitch reference to the boy through use of a pronoun. However, as children got olderthey began (like adults) to use a nominal expression to accomplish the task. Forswitching reference to the dog, children of all ages (like adults) preferred to use anominal expression, not a pronoun. For maintaining reference at a local level, all agegroups (including adults) preferred to do so with a pronoun, regardless of the characterbeing referred to.Bamberg sees the different treatment of the two protagonists as being highly significant.He reasons that it is not possible to explain the highly preferential use of pronouns toswitch reference to the boy as being deictic in nature because of the different method theyoungest children consistently chose to switch reference to the boy, as opposed to thedog. Rather, he explains the phenomenon in terms of the “thematic subject” strategypreviously proposed by Karmiloff-Smith. The pronominal form used in the function ofswitching reference is interpreted to exclusively refer to the boy, chosen to be the “mainprotagonist” of the story. Bamberg believes that, in making this distinction, the childrenare choosing a strategy through the use of pronouns for “advancing thematicprogression” of the narrative (1981: 97), thereby demonstrating knowledge of thenarrative whole. The children appear to be using pronouns when they believe that indoing so they will provide thematic coherence to the whole text; they use nominalswhen they believe thematic progress will be interrupted (i.e. when characters other than16the “main protagonist” are being referred to). Bamberg interprets this strategy ofthematic advancement as a “global level discourse organization device” (1987: 98), ofwhich the “thematic subject” strategy is a subcomponent.After closer analysis of individual stories, Bamberg concludes that children begindevelopment of pronoun anaphora by employing this “global” thematic advancementstrategy, then move through a stage of applying “local” contrastive rules while stillkeeping the thematic strategy at a global level. In other words, as this second strategycomes to be used children are beginning to recognize at a purely local, intersententiallevel that nominal reference to the boy must be inserted if the use of a pronoun couldrefer (locally) to either the boy or the dog. However, at the same time, in unambiguouslocal contexts pronouns will still be used exclusively to switch reference to the boy.During this intermediate stage of acquisition, Bamberg notices that there is no clearform/function relationship which serves to maintain or switch reference; rather there aretwo strategies being applied on different levels. Bamberg therefore proposes that thefinal acquisition of the adult system of pronominal anaphora is the result of the mergingof the local and global strategies to produce one strategy that satisfies both a localdisambiguating and a textual cohesive function.It can be drawn from the discussion above that Bamberg’s interpretation of his datasupport and further explain Karmiloff-Smith’s results (excepting the younger age atwhich he found the “thematic subject” strategy to appear). By viewing the “thematicsubject” strategy as a subcomponent of an early “global” strategy which must later be17blended with “local” rules applied to prevent ambiguous reference, Bamberg is able togo on to propose “forces” which may be driving the child toward the adultlike anaphoricuse of pronouns. First, and perhaps most obvious, children in a literate environment areconstantly confronted with narratives employing the adult system of anaphoric reference.The children are therefore pressured to resolve the differences between the adult inputand their own productions. Second, Bamberg proposes that the emergence of the “local”strategy of contrastive nominalization, which interrupts the “global” strategy at a locallevel, serves to push children toward a resolution of the strategies employed at the twoseparate levels. Therefore, the proposed forces are both external and internal withrespect to children’s developing language system.The combined results of these two studies, in particular the interpretations of the dataset out by Bamberg, lead to interesting conclusions regarding what is presently knownabout the development of pronominal anaphora in narrative production. First, Bambergfound evidence of consistent nondeictic use of pronouns in the productions of childrenas young as 3;6, as well as sensitivity to the thematic development of the story. Thissensitivity to thematic development is consistent with comprehension studies, whichhave pointed to knowledge of an adultlike story schema by three to four years of age(e.g. Hudson & Shapiro, 1991). It is not surprising, and indeed should be expected, thatas children acquire story schema knowledge they should attempt to incorporate thatknowledge into their early narrative productions. Second, while children begin toexhibit this merging of story schema knowledge and linguistic portrayal of thatknowledge at such a young age, their attempts do not consistently reflect the adult18system of pronominal anaphora until after the age of ten years. Rather, there is adevelopmental progression through which children move before the “global” strategy ofthematic advancement and the “local” strategy of disambiguating referents merge tofinally mirror the adult anaphoric strategy, which satisfies both local and textually globalneeds for cohesion and coherence.HypothesesThis examination of the literature pertaining to the nature and development ofpronominal anaphoric reference leads to several hypotheses regarding the proposedstudy. First, it is expected that the results of this study with English-speaking childrenwill be comparable with those found for German-speaking children. Specifically, it isexpected that the data will contain evidence that English-speaking children as young as3;6 are consistently using pronouns in a nondeictic, referential manner in their narratives.Beyond this, if Bamberg’s interpretation of. his own data is valid, it is expected that thedata from English speaking children will support his proposed developmental processin the acquisition of anaphoric pronominal reference, by demonstrating the presence ofa thematic advancement strategy as described above.Second, it is expected that the referential use of pronouns by all children will be affectedby story familiarity. As was discussed above, there is evidence that when children aremore familiar with the structure, content and organization of a story, their use ofcohesive devices is more developmentally advanced. For the most part, such evidencewas obtained through cross-sectional developmental studies, in which older children19(with more developed story schemas) told more linguistically advanced stories (Hudson& Shapiro, 1991). However, the principle will be extended to the context of this study,and the expectation here is that when children have their story knowledge supported (byexposure to an adult model), there will be an improvement in their ability to usecohesive devices to serve specified functions. In particular, the children maydemonstrate more frequent usage of pronominal reference and/or a stronger tendencyto utilize the thematic advancement strategy when telling a familiar, as opposed to anunfamiliar, story. If in fact these children are in the active process of acquiring a storyschema, in addition to acquiring the use of pronominal anaphoric reference, then byfamiliarizing the children with a story (thereby providing a model for story content andstructure) enough of the cognitive burden may be removed from this area of knowledgerequired for narrative production to allow more effort in aspects of the referentialsystem.Third, it may be expected that the results of the proposed study for children youngerthan 3;6 will more closely mirror those found in the study of Karmiloff-Smith in twoways. First, they may demonstrate deictic, nonreferential use of pronouns in the contextof the narrative task and an inability to create textual coherence. This may reflectinsufficiently developed knowledge of a story schema and/or the referential system, oran insufficient ability to integrate these (and other) areas of knowledge required for thestory production task (Hudson & Shapiro, 1991). Second, even if the children evidencereferential use of pronouns, there may be a tendency for clause-initial pronouns to be20reserved for reference to the story protagonist, an integral part of Karmiloff-Smith’sthematic subject strategy.21CHAPTER TWOMETHODOverviewThe design of this study was intended to replicate part of Bamberg’s 1987 study, in orderto allow direct comparison of the results. A second purpose was to compare the resultswith the Karmiloff-Smith’s study (1981). Finally, the study was intended to measure anychange in performance with respect to the use of pronominal reference in a narrativebetween a time when the children were unfamiliar with a story and a time when theywere familiar with it.It was hypothesized that the data of this study of English-speaking 3-year-olds wouldsupport Bamberg’s results for German-speaking children of the same age, namely thatthey would use the proposed thematic advancement strategy. Further, it was expectedthat story familiarity would have a positive effect on the ability of children to apply thisthematic advancement strategy. Finally, it was hypothesized that the children would usepronouns in a manner that reflected the results of Karmiloff-Smith: either they woulduse pronouns in (a) a purely deictic manner, not referentially (in conflict with Bamberg’sresults, but in agreement with Karmiloff-Smith), or (b) referentially, but more closelyreflecting the thematic subject strategy proposed by Karmiloff-Smith.22The study involved analysis of recorded stories elicited from ten 3-year-olds. The studywas conducted in three parts for each subject: (1) unfamiliar condition, (2)familiarization, and (3) familiar condition. The unfamiliar condition involved elicitationand recording of a story told by each child in a private room at her preschool site duringregular preschool hours (with one exception, when elicitation occurred in the child’shome). The story was elicited through presentation of a picture book provided by theresearcher and previously unknown to the child (Frog, Where Are You? [Mayer, 1969]).In the second part, familiarization, the book was sent home with the child for one week,during which time the parent(s)/guardian(s) “read” the book with the child on aminimum of four different occasions. The final stage involved repeat elicitation andrecording of the story told by the child exactly one week following the original storyelicitation. The stories were then transcribed and analyzed for the children’s use ofnouns and pronouns to refer to the two main characters in the story. A picture-by-picture synopsis of the story is presented in Appendix A.SubjectsTen subjects between the ages of 3;O and 3;11’ were selected from the regular aftendersin the preschool program at the U.B.C. Child Study Centre. Subjects were selected withthe assistance of the Program Coordinator of the U.B.C. Child Study Centre, who wasasked to nominate children according to the following criteria:1. between the ages of 3;O and 3;112. monolingual English home environment1 Ages are reported as years;months.days.233. no known physical, mental or emotional handicaps4. no known delays in language acquisitionA parental consent form was sent to the caregivers of eighteen children who met theabove criteria, and consent was given for the participation of sixteen of these children.From these, ten subjects (six girls and four boys) were chosen according to subjectavailability at times allotted for data collection. The ages of subjects at the time of thefirst story elicitation (T1) is listed below (Table 1).In order for chosen subjects to become more familiar with the experimenter beforeleaving their classrooms for the purpose of data collection, the experimenter spent atleast one session in the classroom participating in regular classroom activities andTable 1. Gender and age of subjects at time of first story elicitation (T1)Subject Gender Age at T1* inyears;months.daysSi BT 3;8.12S2 JT 3;5.i7S3 KS 3;8.5S4 SK 3;6.2S5 DA 3;9.i2S6 JB 3;6,2S7DM 3;3.5S8 RL 3;4.22S9 RS 3;6.18SlO EH 3;2.3*Age range: 3;2.3 - 3;9.1224informing subjects that “next time” they would be given the opportunity to go with theexperimenter to play a “story game.”Procedures for Data CollectionPart I (Unfamiliar Condition):The experimenter invited chosen subjects individually to come outside their classroomto read a book with her. If a child refused to participate, even with gentleencouragement, she would be left and approached again at a later time, but was neverrequired to participate against her wishes. No child refused to participate. Each childwas asked to look through the picture book Frog, Where Are You? (Mayer, 1969) until shefelt she knew the book well enough to tell the story. The experimenter asked the childif she was ready to tell the story. If the child responded “yes,” the child and theexperimenter then looked through the book together while the child told the story.Some difficulty was encountered in this part of the data collection process, when a fewsubjects began to tell the story immediately when looking through the book for the firsttime. For certain of these subjects, when asked to tell the story again after becomingfamiliar with the book, they had become disinterested in the task, apparently feelingthey had already finished what the experimenter had asked them to do. For thesesubjects, the first telling of the story is marked in the data. It was decided that the“better” of the two stories would be taken, even if that was the first story (i.e. with noself-familiarization). The justification for this decision and the method for deciding the“better” story are discussed below.25The experimenter provided encouragement and showed interest in the child’s story, butdid not in any way enhance the content or enhance or model the use of pronominalreference in the child’s production. The type of prompting and responding used, whichconsisted of nonpropositional utterances (such as “Mmhmm,” or “wow!”) has been shownnot only to provide contingent responses to the child’s utterances, but also to encouragethe child to keep participating in the narrative task (Peilegrini & Galda, 1990).Nonpropositional utterances of this type were supplemented by repetitions of some childutterances, primarily to assure accurate transcription of unpredictable or difficult-to-hearutterances. Finally, utterances such as “Then what happened?” or “What’s going on?”were used when it was necessary to encourage the child to remain on task. Utteranceslike the latter (e.g. “Then what happened?”) while chosen to be compatable withBamberg’s methods, were also deemed to be appropriate for the elicitation of a story asopposed to other kinds of narrative or picture description. The words were chosencarefully and limited to only those chosen because it has been shown that “instructionsto children [represent] a situational context to the speakers that obligate[s] them to formdistinctions in organizational structure of their language” (Liles, 1993: 872).The child was encouraged to tell the story as many times as she wished. Theexperimenter gave positive feedback regarding the first telling (e.g. “What a good story!”)and asked the child directly if she would like to tell the story again, but did not insiston multiple tellings. The motivation for encouraging the children to tell the story morethan once comes from evidence which suggests that when children are more familiarwith events and the structure in a story, they will be more likely to demonstrate the26upper limits of their knowledge of narrative cohesion (including pronominal reference)(Hudson & Shapiro, 1991). Therefore, by allowing the child some kind of “self-familiarization” process in the first part of this study, it was intended that the childrenwould have the opportunity to tell the most linguistically sophisticated story they couldwithout yet having the benefit of an adult model. Only three subjects actually ventureda second story in the first telling, apparently for the same reasons previously mentioned.For subjects who did give two stories, the “better” of the stories was chosen for analysis,according to completeness of the story with respect to six basic story schema elements:(a) Initial Event Chain (two parts), (b) Search Motif, (c) End (three parts) (based onBerman, 1988; see Transcription & Coding section below for details). Therefore, even iftwo stories were given, the second was not necessarily the one chosen for analysis.The story was recorded using both audio and visual recording equipment. However,the recording of the first telling by Subject 9 (RS) was not recorded by video due tobattery failure, and neither telling of Subject 10 (EH) was recorded with video due to herdistractibility. Audio recording equipment included the following: audio recorder(Marantz, model PMD42O), microphone (Samson remote), transmitter (Samson VHF FMST-2) and receiver (Samson VHF FM SR-2). Video recording equipment included: camera(JVC Super VHS CF-S 550), tripod (Davis & Sanford Model F-lU) and tapes (ScotchColour Plus High Grade). Video tapes were viewed on a Mitsubishi Hi-Fi U54 videocassette recorder. Transcription of audiotaped material was done over headphones witha General Electric 3-526OA audio cassette player.27Part II (Familiarization):During the week following the first story elicitation, a copy of Frog, Where Are You? wassent home with the child. The parents were instructed (by letter) to “read” the story tothe child on at least four different occasions throughout the week. The parents wereinformed of this aspect of the study in the letter of initial contact, but a letter ofreminder was sent home with the book, as well as reinstruction that any questionsregarding procedure or the study itself could be directed to the experimenter or herfaculty supervisor. The parents were also asked to record one example of their storiesduring that week, for possible later analysis.Part ifi (Familiar Condition):One week after the first story elicitation (Part I), the researcher again invited theindividual children to tell the same story. As in Part I, each child was asked to tell thestory while looking through the picture book Frog, Where Are You? with theexperimenter. Again, the child was encouraged to repeat telling the story if she wished.In the second telling only two subjects gave more than one story, and in these two casesthe “better” of the stories was taken for analysis according to the six story elementcriteria noted in Part I above. All experimenter prompts and responses during thechild’s telling of the story were restricted as outlined in Part I above. This second tellingof the story was also recorded using audio and video recording equipment (except forSubject 10, as noted in Part I).28Transcription & CodingTranscription of raw data was done on a clause by clause basis, following the guidelinesof Berman et al. (1986), who define a clause as a “unified predicate,” i.e. one which“expresses a single situation (activity, event, state)” (p. 37).The coding procedure employed conformed to that used by Bamberg (1987) to the extentthat the differences between the English and German languages allow. All aspects ofBamberg’s coding were not explicitly detailed in his study; where his explanation ofcoding procedures was vague, coding was done by comparison with examples given inhis text.Form CategoriesAll mentions of the two main characters of the story (the boy and the dog) were firstcoded according to their form and function. Coding categories for form were dividedinto three categories:1. Nominal: reference to a character by a full nominal expressione.g. The boy looked in his boot.2. Pronominal: reference to a character by a personal pronoune.g. landed on a reindeer.3. Ellipsis: referential subject of clause not explicitly mentioned.e.g. And look.— calling “Freddy, Freddy!” in the hole.29Function CategoriesThe two categories for function were:1. Switch Reference (SR): Introduction of a character into the story, orreintroduction of a character after mention of other characters or events.2. Maintain Reference (MR): Continued mention of a character after his introductionor reintroduction, with no interruption by mention of other characters or events.SubcategoriesEach clause coded according to the form and function categories above was furthercoded according to whether it fit one or more of the following subcategories:1. Joint reference to the boy and the doge.g. They looked everywhere...2. Proper noune.g. (urn) well Andrew looks in the boots.(reference to the boy)3. Maintain reference to last-mentioned character by full nominal when thisreference occurs at the outset of a new event or page4. “Appropriate” use of a pronoun to switch reference (SRi) where a full nominalmay otherwise be expected (see below for explanation of “appropriate”)e.g. He chased that dog. -> he (SR) -> that dog (SR)He trying to chase the dog. -> he (SRi) -> the dog (SR)In the example of subcategory 4 above, the second mention of the deer is technically apronoun serving a switch reference function to a character other than the protagonist in30the story. Because the second clause contains same subject and object in identicalposititions, and actually repeats the information in the first clause, there is no doubt asto the referent of the pronoun in the second clause. In fact, had the second clause read“He trying to chase him” the object pronoun “him” would also have been coded“appropriate” due again to the utter clarity of the referent of the pronoun. Althoughsuch use of pronouns could be construed as a violation of the Bamberg’s developmentaltheory of the acquisition of pronominal anaphora, because of the lack of ambiguity ofthe referent of the pronoun in examples such as these, they were deemed “appropriate”uses of the pronoun in the switch reference function (as will be discussed in chapter 3).The coding of reference according to the four subcategories described above allows fordiscussion of the children’s use of pronominal vs. nominal reference forms, but does notaffect the total counts of pronominal and nominal reference. Therefore, the data are stilldirectly comparable to Bamberg’s.Mention of Other CharactersThe mention of characters (and objects) other than the boy or the dog was also coded,but for a different purpose. The purpose of this coding (which also recorded whetherthe expression was nominal or pronominal in form) was to determine if the mention ofthis other character was important in the thematic advancement of the story. If it wasconsidered to be so, then the next mention of one of the two main characters was codedas a ‘switch reference’. If mention of the other character was not considered to beimportant to thematic development, then reference to a main character immediately wasconsidered a ‘maintain reference’ if and only if this character was the object of reference31immediately prior to the “interrupting utterances. Intervening mentions of othercharacters judged not to be important for thematic advancement were primarilydescriptive statements (e.g. The moon is out) and narrator comments (e.g. What’s that?or He’s funny!), Mentions of other characters who were judged to be part of the thematicadvancement of the story included the telling of episodes which were part of but notcentral to the main story line (e.g. the episode of the bees chasing the dog) or characterswhich interacted with one or both of the main characters or participated in the mainstory line (e.g. the frog’s actions at the beginning of the story, or the episode in whichthe deer carries the boy to the cliff-side). The following examples show 1. a case whereintervening utterances are not important to thematic advancement, and 2. a case wherethe intervening utterances are important.1. The frog is in. ->the frog (SR)The moon is up. [static observation, of no import to thematicdevelopment]Oh, he’s out! ->he (MR) - frog2. He landed on a reindeer ->he (SR)- boy->a reindeer (SR)Then the reindeer throws him down into the river->the reindeer (MR)->him (SR)- boyThe referents and function (i.e. switch or maintain reference) of pronouns in clause-intialposition were also coded to allow comparison with Karmioff-Smith’s data regarding32positional constraints on referential pronoun use by young chidren (1981). Clauseswhich had been identified as important to the thematic advancement of the story weremarked if they contained an initial referential pronoun. These occurrences of clause-initial pronouns were then broken down with respect to their reference function, theirreferent (boy or other) and, in the case of the maintain reference function, with respectto whether or not the occurrence of maintain reference was “local.” “Local” maintainreference pertained to a clause which contained a pronoun (serving the maintainreference function) which refers to the character most recently established by a fullnominal. Either no clauses intervened between the (re)introduction of the character bya nominal, or any intervening clauses contained reference by pronouns to the only thesame character and no reference of any type to any other character. An example of“local” maintain reference is as follows:There’s a big big rock. (SR - rock)biting his knee. (MR (local) - rock)Mention of Characters Accompanied by Finger PointTo allow further comparison with Karmiloff-Smith’s data and results (1981), each use ofa pronominal form for mention of a character was coded according to whether it wasaccompanied by a physical point. Data showing co-occurrence of pronominal forms andphysical deictic point were used in analysis designed to indicate possible deictic asopposed to referential use of pronouns.33Story Structure ElementsEach story was analyzed for basic elements that reflect the developmental sophisticationof the child’s knowledge of the story schema, as demonstrated in this production task.This general analysis of story organization and structure provided a frame for discussionof children’s ability to achieve global coherence and maintain cohesive structure throughlocal use of cohesive devices (in this case pronominal reference). To this end, each storywas coded and analyzed for the following:First, the stories were coded as including or not including the following six basic storyelements (based on Berman, 1988)2:a. Initial Event Chain: [Onset of Problem]1. Frog leaves jar2. Protagonists(s) discover that frog has goneb. Search motif: [Goal]3. Protagonist(s) search for missing frogc. End: [Resolution of Problem]4. Boy finds frog5. Boy takes a frog6. Frog is same as or substitute for missing pet.Indicators of Picture Description or Story LineThe stories were also coded for forms that can differentiate between the case in whicha child is treating each page of the story book as a separate picture description task, and2 These six story elements were also used as the criteria for judging the “goodnes&’of a given story in order to choose between two stories for the purpose of codingand analysis (see Procedures for Data Collection).34the case in which a child recognizes a story line that results from the sequentialinterconnectedness of characters and events represented in the pictures.These forms include the following:1. Indicators of picture descriptiona. The use of such forms as Here’s/there’s a..., or I see a... at thebeginning of or during a set of utterances pertaining to each pictureframe;b. The use of the indefinite article a in the first mention of a characterfor each picture frame, regardless of prior appearance or mention ofthat character in previous picture frames.2. Indicators of story linea. The use of the definite article the in mention of characters whichhave received prior mention;b. Use of temporal forms such as then, next or after that, as well asother connectives such as so, because, and and but, all of whichindicate the continuity and interconnectedness of subsequent events.Words Per ClauseClauses that contributed to thematic development were also marked for number ofwords per clause in order to allow calculation of the mean length of clause (MLC) foreach story.35AnalysisA prime motivation for the arrangement and presentation of the data is directcomparability with Bamberg’s results. To achieve this end I will describe the use ofanaphoric reference by individual subjects and the group as a whole, which comparesperformance across subjects. Similarly, the results of this study will be arranged to allowcomparison with Karmiloff-Smith’s, in particular data pertaining to the deictic use ofpronouns and evidence of local cohesion and/or textual coherence in the stories.In addition, data will be presented which shows whether and how the data “splits” withrespect to age, i.e. “young” vs. “old” 3-year-olds and their use of pronouns to makereference. To further explore the notion of a developmental “time-line” for language(including pronominal anaphora), the various developmental measures utilized in thisstudy will be checked for correlation. Analysis of errors and spontaneous self-corrections in individual stories will be discussed. Finally, data and informationcomparing the children’s performance on many of the above-listed measures in theunfamiliar and familiar conditions will be presented, to assess the effect of storyfamiliarity on the use of pronominal reference.Due in most part to the number of different categories into which instances of referencewere coded, data for individual children did not produce large enough sample sizes percoding category to allow statistical comparison between categories describing use ofpronouns in making reference. Individual data will be discussed descriptively.However, data summed for the group did result in sample sizes sufficient to allow36parametric statistical procedures to be applied. Therefore, the significance of differencesin group performance with respect to various measures could be assessed. With fewexceptions, the hypothesis tests involved in this study concerned comparison of singlepopulation proportions to preselected null values, and comparison of two populationproportions. The appropriate distribution to describe such random variables is thebinomial distribution, which is usually well approximated by the normal distribution inmany cases, including all cases involved in this study. This leads appropriately to theuse of the standard normal distribution, or z distribution, in the hypothesis testsconcerning proportions in this study (Triola, 1992).The only hypothesis test in this study which did not concern population proportionswere tests of correlation between selected pairs of measures. The Pearson correlationcoefficient, r, was chosen as the appropriate test statistic in these hypothesis tests (Triola,1992).Throughout, hypothesis tests using the z distribution were applied with a consistentlevel of significance (cc) at 0.05. The significance level is deemed appropriate for a studyof this type, which aims to replicate results not previously tested for significance.Description of the statistical treatment of various measures will be discussed in detailbelow with respect to each of the hypothesis statements set out in chapter 1.37Hypothesis 1: The use of pronominal anaphoric reference by English-speakingchildren will compare with that of German-speaking children, and in doing sosupport the developmental process proposed by Bamberg (1986). Specifically, 3-year-olds will use a global thematic advancement strategy in their narratives.As discussed in chapter 1, Bamberg’s proposed theory of the developmental process forthe acquisition of the adultlike use of pronominal anaphoric reference contains fourstages, beginning with deictic, nonreferential use of pronouns and ending with the adultpreferences for referential pronoun use. He proposed that children as young as 3;6 areable to use pronouns in a referential (as opposed to deictic) manner and show the useof a “global” thematic advancement strategy, whereby they favour the use of a personalpronoun to introduce and reintroduce the story character deemed to be the mainprotagonist. Children would concomittantly favour the use of a full nominal tointroduce and reintroduce characters other than the main protagonist. Pronouns remain,however, the form by which reference to these characters is maintained.In order to test this strategy against the data collected here, all instances of anaphoricreference to the two main characters in the story (the boy and the dog) were recordedand coded according to their form (pronominal, full nominal, or ellipsis) and theirfunction (switch reference or maintain reference). This procedure of coding was donein a manner as similar as possible to that in Bamberg’s study. The clauses were codedand tabulated at T1 (unfamiliar condition) and T2 (familiar condition).38The two characters (boy and dog) are appropriately chosen because they both occurthroughout the entirety of the story, and the boy is the natural protagonist of the story(although the dog is part of and sometimes central to certain episodes). Therefore, if infact the children selected the boy as the story protagonist, then their preference for usingpronouns to make anaphoric reference to the boy in contrast with nominal reference tothe dog could be analyzed with respect to Bamberg’s developmental progression.Further, for a given character, preference for use of pronouns as opposed to fullnominals to serve different reference functions could be assessed. In all statistical tests,the null hypothesis and alternate hypothesis were set up such that if the data showedsignificant trends similar to those consistent with Bamberg’s theory, then the nullhypothesis would be rejected in favour of the alternate.The summed counts of anaphoric reference forms and functions for all subjects providedthe data base to compare these English-speaking children’s use of pronominal referenceagainst Bamberg’s theory of pronominal reference development. Observation of trendsin the data (namely distribution of nominal vs. pronominal forms, and the comparativenumber of total references to the boy vs. the dog) was made to provide an initialcomparison of my data with Bamberg’s. Following this initial check for datacomparability, two different tests were performed which looked at (1) the significanceof any difference in the proportions of pronominal and nominal reference used to switchor maintain reference to the boy and to the dog, and (2) the significance of anydifference in the proportion of pronouns used to make a given type of reference to theboy compared with the dog. The first test would therefore show any tendencies in how39the children are using full nominals and pronouns to serve different reference functionsfor the protagonist, and for another (main) character. The second test would indicateany tendency that exists for the children to use a different strategy for referring to theprotagonist and nonprotagonist. These two tests are sufficient to compare the obtaineddata with Bamberg’s according to hypothesis I above.Hypothesis 2: The referential use of pronouns by all children will be affected bystory familiarity. Specifically, if the children are in the developmental process ofacquiring pronominal anaphoric reference, then when they are more familiar withthe content and schema of a story, they will exhibit more developmentallyadvanced use of pronominal reference, and in addition may exhibit a greaterquantity of pronominal reference.Evidence to support or not support this claim requires information regarding thedifference in performance between the unfamiliar condition (T1) and familiar condition(T2) on the same measures that were used to test the fit of the data to Bamberg’s theoryof pronominal reference development. In other words, the goal was to determinewhether or not the children showed a significantly greater tendency to use pronouns inthe manner proposed by Bamberg for their age range (i.e. the use of the “global”thematic advancement strategy). Therefore, tests were designed to compare (I) thechildren’s preference to use pronouns to switch reference to the boy at T1 and T2, and40(2) any difference between T1 and T2 in the children’s preference to use pronouns toswitch and maintain reference to each of the boy and the dog.The possibility exists that individual children may have utilized strategies for makingreference that deviated from those used by most other children, but not enough to altergrouped data. This issue will be addressed in the discussion of individual data.Also of interest here was evidence of whether children are “better” at telling the storyafter having received an adult model, therefore lending support to the notion thatfamiliarity may ease their “cognitive load” with respect to the content and schema of thestory, allowing them to demonstrate more mature use of pronominal reference.First, the “goodness” of children’s stories according to the presence or absence of the sixbasic story schema elements at T1 and T2 will be presented. Second, the Mean Lengthof Clause for each subject at T1 and T2 will be presented. Third, the proportion ofclauses which contributed to thematic development out of all clauses in the story(summed for all subjects) will be presented and statistically examined for differencebetween the unfamiliar and familiar conditions.41Hypothesis 3a: The children are using pronouns deictically (as opposed toreferentially) in the story.Hypothesis 3b: The children are using pronouns in a referential manner, but theiruse may more closely conform to the thematic subject strategy proposed byKarmiloff-Smith.In order to determine whether pronouns were being used deictically, rather thanreferentially, data regarding the accompaniment of pronouns by point was investigated.(Simultaneous point with a pronoun is one indicator of deixis, discussed in chapter 1).It was reasoned that if pronouns were being used deictically, then there should exist thetendency for them to be accompanied by a point. This use of point would be adistinguishing mark which would separate pronouns from a nondeictic term, forexample a full nominal. Therefore, a test was designed which compared the number ofpronouns accompanied by point with the number of full nominals accompanied bypoint, considering forms that made mention to the boy and/or the dog in clausesdeemed to be contributing to the thematic advancement of the story (i.e. coded clauses).The goal of this test was to determine if there was any difference in the frequency ofpoint with the pronouns as opposed to the full nominals. If the results showed thatthere was no difference in the distribution of point between the two forms, this wouldpoint to nondeictic use of pronouns.42Further information regarding the quantity and variety of deictic terms in the stories andthe presence or absence of point with these terms was tabulated for individual subjectsand summed for all subjects. These data were not subjected to statistical measures, butwere compared with the data regarding the accompaniment of pronouns and nominalsby point. These results will be discussed descriptively.Karmiloff-Smith found evidence of a thematic subject strategy, but only in children overthe age of six (years older than the children studied by Bamberg). In her description ofthe thematic subject strategy, the utterance-initial position was almost exclusivelyreserved for pronominal reference to the protagonist. Therefore, occurrences ofpronominal reference in clause-initial position were extracted from the transcripts andassessed in order to further compare with the proposed thematic subject strategy byKarmiloff-Smith. Information regarding the utterance-initial pronouns serving anapparent reference function (to any character) in coded clauses will be presented. Thesetables will show the percentage of initial, subject pronouns which had the boy asreferent, along with the reference functions of these pronouns.A secondary interest in this study was to determine whether measures of languagedevelopment other than pronominal reference were correlated in this narrative context.If two measures are correlated then one might serve as an indicator of the other. Twoof the measures used to assess whether the children were “better” at telling the story inthe familiar condition (story “goodness” and MLC) were also selected as measures oflanguage development, in addition to chronological age. Such use of MLC and the43presence of story schema elements in narratives as developmental scales are welldocumented in the literature (e.g. Brown, 1973; Stein & Glenn, 1982). A correlation testbetween each pairing of developmental measures was therefore carried out.Closely related to the test of correlation between the developmental measures, it was ofinterest to see whether there was any relationship between chronological age and the useof pronouns to make anaphoric reference. Such a result would support the proposeddevelopmental sequence for acquisition pronominal reference proposed by both Bambergand Karmiloff-Smith. Therefore, a test was designed to determine any differencebetween the youngest four subjects’ and oldest four subjects’ preferential use ofpronouns to switch reference to the boy, anticipating that the older children would showa greater preference of pronouns to fill this reference function.Finally, throughout the presentation and discussion of the results, instances wherereference to the boy and/or reference to the dog are in question for analysis, there willalways be two different versions of the data and results presented, one which includesjoint reference to the boy and the dog (for example by pronouns such as “they” or“them”) and one which does not include joint reference. During the coding procedure,instances of joint reference were kept separate, and assigned to both the boy and the dog(“double coded”). Bamberg did not describe the case of joint reference to the twocharacters, so it is unknown how he dealt with these cases. The results are presentedboth ways in the interest of precision and completeness. Any differences in the resultsfor the cases of included and excluded joint reference will be discussed.44CHAPTER THREERESULTS AND DISCUSSIONOverviewIn this chapter I will present and discuss the results of this study. First, the results ofthe analyses of data grouped for all subjects will be presented and discussed in the orderof the three hypotheses set in chapter 1. This will be followed by the results andaccompanying discussion with respect to across-subject measures. I will then presentand discuss relevant errors and self-corrections made by subjects, which give furtherweight to conclusions already drawn from the data. Finally, I will present and discussthe analysis of individual stories.Group DataComparison with Bamberg’s Data and ResultsThe raw counts and the percentage of total reference for each pairing of the categoriesof reference form and function are shown in tables 2a and 2b. Table 2a shows allreference occurrences, including joint reference to the boy and dog. Table 2b shows thesame information as table 2a, but excludes joint reference to the boy and dog. (For abreakdown of the form/function category pairings into subcategories, please seeAppendix C.)45Table 2a. Number (and percentage) of nominal, pronominal and elliptical reference tothe boy and to the dog, at T1 and T2 (joint reference to boy and dog included)SWITCH MAINTMN SUMT1 T2 T1 T2 T1 T2(%) (%) (%) (%) (%) (%)To the boy:Nominal 14 (13) 10 (6) -- -- 1 (1) 14(13) 11 (7)Pronominal 56 (50) 88 (57) 35(31) 42(28) 91(81) 131 (85)Ellipsis 3 (2) 4 (2) 4 (4) 9 (6) 7 (6) 13 (8)Sum 73 (65) 102 (65) 39(35) 53 (35) 112 155To the dog:Nominal 40 (42) 24 (21) ---- 3 (3) 41(42) 27 (24)Pronominal 25 (27) 40 (36) 26(28) 41 (37) 51(55) 81 (73)Ellipsis-- -- -- -- 3 (3) 3 (3) 3 (3) 3 (3)Sum 65 (69) 64 (58) 29(20) 47 (43) 94 111Sum to the boy 138 166 68 100 206 266&_dog:46Table 2b. Number (and percentage) of nominal, pronominal and elliptical reference tothe boy and to the dog, at T1 and T2 (joint reference to boy and dog excluded)SWITCH MAINTAIN SUMT1 T2 T1 T2 T1 T2(%) (%) (%) (%) (%) (%)To the boy:Nominal 13 (14) 10 (8) -- -- 1 (1) 13(14) 11 (9)Pronominal 46 (50) 67 (53) 26(28) 36(29) 72(78) 103(82)Ellipsis 3 (3) 4 (3) 4 (5) 8 (6) 7 (8) 12 (9)Sum 62 (67) 81 (64) 30(33) 45(35) 92 126To the dog:Nominal 39 (53) 24 (30) -- -- 3 (3) 41(42) 27 (24)Pronominal 15 (20) 19 (23) 26(28) 41 (37) 51(55) 81 (73)Ellipsis -- -- ---- 3 (3) 3 (3) 3 (3) 3 (3)Sum 54 (73) 43 (53) 20(27) 38 (47) 74 81Sum to the boy 116 124 50 83 166 201&_dog:47Distribution of Reference Between the Two Main CharactersTo compare the number of references to the boy with number of references to the dog,it is most telling to look at table 2b, which excludes joint reference to both characters.This eliminates any question of referent in a joint reference situation. The summed datashow that, out of the 166 references exclusively to the boy or the dog, 92 (or 55%)referred to the boy, and 74 (45%) referred to the dog at T1. At T2, 61% of the 207references were exclusively to the boy and 39% to the dog. The trend in the data to referto the boy more often than to the dog is consistent with Bamberg’s data, which alsoshowed that, although the boy and the dog are depicted in approximately the samenumber of pictures, there is a greater tendency to refer to the boy. This may reflect andsupport the children’s apparent decision to choose the boy as the protagonist of thestory. From Bamberg’s results, it was expected that more of the children’s referringexpressions would be directed toward the protagonist of the story.Distribution of Nominal and Pronominal FormsThe distribution of use of pronominal vs. nominal forms to refer to the boy and to thedog irrespective of function can be seen in table 2a (including joint reference) and table2b (excluding joint reference). With respect to the boy, the data show the children’sstrong preference to use pronouns to refer to the boy at both times, regardless of theinclusion of joint reference (81% at T1 and 85% at T2 when joint reference is included,78% at T1 and 82% at T2 when joint reference is excluded). This is consistent withBamberg’s data, which also showed that children in the 3;6 - 4 year age groupdemonstrated a preference for referring to the boy with a pronoun. In fact, the data in48this study show a greater trend toward preference of the pronominal form for referenceto the boy (Bamberg found only 53% of references to the boy were pronominal).For reference to the dog, the data did not show a clear preference to use nominal orpronominal forms (see tables 2 and 3). In the case where joint reference to the boy anddog is excluded (table 2b), at T1 a slight preference for the nominal form was shown(53%), but at T2 there was a preference for the pronominal expression (64%). Bambergfound a preference in referring to the dog to use nominal expressions (61%).Of importance to note is the ratio of the number of expressions that switch reference tothe dog to the number of expressions that maintain reference to the dog. In Bamberg’sdata, similar to the T1 numbers listed above for this study, there were approximatelytwice the number of switch reference as maintain reference expressions in reference tothe dog. However, in the T2 condition above, the ratio of switch to maintain was almosteven (43:38). This difference is noted, but is of no real significance when the tworeference functions are discussed separately.Of note as well is the more extreme preference for use of pronominals for reference tothe dog when joint reference to the boy and dog is included (table 2a). In this case, atT1 the percentage of pronominals was 55% (a switch of preference), and at T2 it was 73%.Important here is the fact that all joint references to the boy and dog (regardless offunction) are pronominal. This may indicate that when the boy and dog are referred tojointly, the children may be adhering to the strategy for exclusive reference to the boy49(which utilizes pronouns to switch reference as will be discussed below), overriding thestrategy for referring to the dog, despite the dog’s equal role as referent.The results obtained above indicated that, while there were differences from theBamberg’s results as noted, these differences did not seriously affect the basic similarityof the two data sets. Therefore, closer and more specific analysis of the data waswarranted.The next analytic step was to investigate the distribution of form with respect to functionfor the boy and for the dog. The differences between pairings of form and function fora given character, and the difference in choice of form to serve a given function betweenthe two characters were measured. Going beyond Bamberg’s study, the data weresubjected to statistical tests, such that the trends in the data could be reported with aknown degree of confidence.Switch ReferenceWith respect to Bamberg’s proposed theory, it was expected that children of three to fouryears would prefer to use pronouns to switch reference to the protagonist (boy). Thispredicts that the proportion of nominals out of the total number of reference forms usedto switch reference to the boy should be less than half (*H1 in tables 3a and 3b). As canbe seen in tables 3a and 3b (including and excluding joint reference, respectively) at bothtimes there was evidence that nominals formed significantly fewer than half of the formsused to switch reference to the boy.50Table 3a. Proportions of reference in nominal form made to each main character (boyand dog) for specified functions at each time (joint reference to boy and dog included)x n test z Reject if: Rejectstatistic H0?(z)Switch 14 70 35 -5.02 -1.64 z<zBoy, T1Switch 10 98 49 -7.88 *‘j’Boy, T2Switch 40 65 32.5 1.86 +1.64 z>z #YDog, T1Switch 24 64 32 -2.00 #NDog,_T2Maintain 0 35 17.5 -5.92 -1.64 z<zBoy, T1Maintain 1 44 22 -6.33Boy, T2Maintain 0 26 13 -5.10 *Dog, T1Maintain 3 44 22 -5.73 *\/Dog, T2cc U.U5x = number of nominal reference to a specified character in a specified function at agiven time.n = summed number of pronominal and nominal reference to a specified character ata given time.= np, where p is the test proportion consistent with the null hypothesis* H0: p = .5H1 : p <.5# H0 : p = .5H1 : p> .551Table 3b. Proportions of reference in nominal form made to each main character (boyand dog) for specified functions at each time (joint reference to boy and dog excluded)x n test Zcrit Reject if: Rejectstatistic H0?(z)Switch 13 59 29.5 -4.30 -1.64 z<z *yBoy, T1Switch 10 77 38.5 -6.50 H *Boy, T2Switch 39 54 27 3.26 +1.64 Z>Zcrjt #YDog, T1Switch 24 43 21.5 0.762 #NDog,_T2Maintain -- 26 13 -4.71 -1.64 z<zBoy, T1Maintain 1 37 18.5 575 11 *yBoy, T2Maintain -- 17 8.5 -3.64 HDog, T1Maintain 3 36 18 -5.00 *Dog,_T2cx U.Ux = number of nominal reference to a specified character in a specified function at agiven time.n = summed number of pronominal and nominal reference to a specified character ata given time.= np, where p is the test proportion consistent with the null hypothesis* ]J: p = .5H1 : p < .5# H0 : p .5H1 : p> .552Therefore, it follows that significantly more than half of these expressions werepronominal in form, supporting Bamberg’s theory of preferred use of the pronoun torefer to the chosen protagonist of the story. This held true both when joint referencewas included (table 3a) and when it was excluded (table 3b).For switching reference to the dog, Bamberg’s theory predicted that the preferential formwould be a full nominal. Therefore, one would expect that significantly more than halfof the forms used to switch reference to the dog would be nominal. This was true forthe first telling of the story (T1), but no evidence for the preference of nominals to switchreference to the dog was found in the second telling (T2), even when the joint referenceto the boy and dog was excluded (although in the latter case more than half of theswitch references were nominal, but not enough to be significant at an a. level of 0.05).This lack of clear preference to use nominals at T2 is inconsistent with Bamberg’s results.The preference for nominal use in switching reference to the dog at T1 is consistent,however, and supports Bamberg’s theory.These results indicate that, in the stories of the subjects in this study, the Englishspeaking children did not make as clear a distinction between use of nominals andpronominals to switch reference to the dog as the German-speaking 3-year-olds inBamberg’s study did. This may be due to the fact that the average age of children waslower for this study than for Bamberg’s (the youngest child from whom he obtained datawas 3;6).53The younger age range may be important, in that the children in this study may be (asa group) at a point in development of this “global” thematic advancement strategy whichis behind that of the children in Bamberg’s study. The question which then remainsimportant is: while the trend toward use of nominals to switch reference to the dog isonly statistically significant at T1, is there still a significant difference at both timesbetween the form/function pairings in making reference to the boy as opposed to thedog? As will be seen below (tables 4a, 4b and subsequent discussion), this significantdifference is found.Maintain ReferenceNext to be looked at was the maintain reference function and the forms used to fill thisfunction for the boy and the dog. According to Bamberg’s results, it was expected thatthe children would choose to use pronominal forms to maintain reference to allcharacters. In other words, once a character was established or reestablished by a switchreference, the children should have used pronouns to continue to make further reference,regardless of the character involved. Therefore, for both the boy and the dog, it wasexpected that there would be a strong preference to use pronominals as opposed tonominals to serve this function. As can be seen in tables 3a and 3b, the proportion ofnominal forms used to serve the maintain reference function was significantly less thanhalf. This is consistent with expectation and Bamberg’s proposed theory.54Reference to the Boy Contrasted with Reference to the Dog.The second set of statistical measures had the purpose of examining any differences inchoice of form to serve a given function for the boy as opposed to the dog. The resultsof the first statistical measure (above) allowed comment as to whether there was adifference between forms preferred to switch vs. to maintain reference to a givencharacter. It did not allow comment as to whether the characters were being treateddifferently with respect to a given function.From Bamberg’s theory, the first expectation from the data is that there should be asignificantly larger proportion of pronouns used to switch reference to the boy asopposed to the dog. Tables 4a and 4b show the results of this measure (joint referenceincluded and excluded, respectively), which support this expectation. This held true forthe data regardless of the inclusion or exclusion of joint reference.As discussed above, there was no evidence for children’s preference at T2 to use fullnominals to switch reference to the dog. However, at T2 significantly more pronounswere used to switch reference to the boy. Therefore, the data still support the notionthat the two characters are being treated differently at T2, even though clear evidence forthe preference of nominals to switch reference to the dog was not found.The second expectation in looking at the reference forms referring to the boy and thedog was that there should be no difference in the proportion of nominal (and thereforepronominal) forms used to maintain reference to both characters. This also held true for55Table 4a. Comparison of proportion of pronominal reference out of total reference madefor a specified function and time, reference to boy vs. reference to dog (joint referenceto boy and dog included)Xb/fld Xd/fld test statistic z Reject H0 Reject H0,(z) if:Switch 0.778 0.385 4.67 1.64 z>z *y’T1Switch 0.863 0.625 3.55 *yT2Maintain 0.897 0.897 0.00 #NT1Maintain 0.811 0.872 -0.830 ‘I #NT2c = U.U5x = number of pronominal reference for a specified function and timen = summed number of all reference types for a specified function and timex/n = sample proportion (p)* H0 : Pb - Pd = 0H1 : Pb- Pd> 0# H0 : Pb - Pd = 0 reject H0 if z> +ZcrjtH1 : Pb - Pd <> 0 or z< Zcrit56Table 4b. Proportion of pronominal reference out of total reference made for a specifiedfunction and time, reference to boy vs. reference to dog (joint reference to boy and dogexcluded)Xb/flb Xd/fld test statistic Zcrjt Reject H0 Reject(z) if: HQ?Switch 0.754 0.278 5.10 1.64 z>z *yT1Switch 0.827 0.441 4.43 *‘T2Maintain 0.867 0.850 0.166 +1.96, z>+z #NT1 -1.96 or z<-zMaintain 0.800 0.868 -0.829 #NT2:z 0.05x = number of pronominal reference for a specified function and timen = summed number of all reference types for a specified function and timex/n = sample proportion (Pea)* H0 : Pb - Pd = 0H1 : Pb - Pd> 0Pb-Pd = 0 rejectHifz> +ZcrjtH1 : Pb- Pd <> 0 or z< -z57the data, with and without joint reference (see tables 4a and 4b). Therefore the datasupport the expectation that there should be no difference in how children treat theprotagonist and another (main) character while maintaining reference.SummaryConsistent with Bamberg’s results, the English-speaking children referred to the boymore often than to the dog, and even more strongly preferred to use pronouns to referto the boy. Inconsistent with Bamberg’s results, the children did not show a strongpreference for referring to the dog with nominal forms. The children used significantlymore pronouns than nouns to switch reference to the dog (consistent with Bamberg’sresults), but only showed a significant preference to use nominals to switch reference tothe dog at T1 (therefore the results were inconsistent with Bamberg’s at T2). Despite this,the children still showed a significantly greater tendency to use pronouns to switchreference to the boy than to the dog, therefore indicating that they were treating the twocharacters differently when choosing referring expressions. Consistent with Bamberg’sresults, the children significantly preferred pronouns to maintain reference to bothcharacters. The results of this section therefore confirm Hypothesis 1 (with the fewnoted exceptions).58Comparison of Performance at T1 and T,Measures to Assess Form/Function Pairings at T1 vs. T2The next set of measurements was designed to determine whether and how performancechanged when the children were more familiar with the story, having received an adultmodel of the story on at least four occasions between the first and second telling. Ofparticular interest here were the measures reported above which tested the children’spreference to choose a pronoun to serve a given reference function for each character.As discussed in chapter 2, the expectation was that if performance did change betweenthe two times, then there should be more evidence of the thematic advancement strategyat T2 (although individual differences in strategy may have been present, and will bediscussed with the individual data). Specifically, at T2, (1) more pronouns would bechosen to switch reference to the boy; (2) fewer pronominals should be chosen to switchreference to the dog, reflecting a further separation in the treatment of the twocharacters; and (3) children should continue to exclusively use pronouns to maintainreference to both characters.Before entering into a discussion of the results, it is important to note that researchwhich has compared the narratives resulting from story generation and story retellingshows that the stories obtained by story retelling are longer and more complete (i.e.contain more information about events) but that the basic level of organization (i.e. storyschema) is not significantly different between tasks (Merritt & Liles, 1987; Ripich &59Griffin, 1988, as cited in Liles, 1993). The two conditions in this study can be viewed asstory generation (unfamiliar) and story retelling (familiar). Therefore, the story datafrom the two tellings can be compared with confidence.First, the proportion of pronominal reference used to switch reference to the boy wasexamined for any change between T1 and T2. This is a key element of the thematicadvancement strategy, and therefore should reflect any change in children’s applicationof the strategy to their story telling. As can be seen in Tables 5a and 5b (joint referenceto boy and dog included .and excluded, respectively), although the proportion ofpronominals used to switch reference to the boy did increase between the two times, thedifference was not significant at the 0.05 level.The other key element in Bamberg’s proposed “global” thematic advancement strategyis children’s preferred use of nominals to switch reference to the dog. However, in thisset of data, the children chose a greater number of pronouns to switch reference to thedog at T2, even when joint reference to the boy and dog were excluded (see tables 5aand 5b). Therefore there was no evidence that the children made significantperformance gains in applying the “global” thematic advancement strategy of choosingpronouns to switch reference to the protagonist and nominals to switch reference toother characters. In fact, the children showed a weaker separation of the treatment ofthe two characters in the switch reference function. It must be kept in mind, however,that there was still a significant difference between treatment of the boy and treatment60Table 5a. Comparison of pronominal reference to the boy and to the dog, for specifiedreference functions, T1 vs. T2 (joint reference to boy and dog included)x/n= p test statistic z Reject H0 Reject(z) if: H?Switch 0.388 -2.69 -1.64 z<z YBoyMaintain 0.448 -0.919 NBoySwitch+ 0.410 -2.68 YMaintainBoySwitch 0.385 -1.86 YDogMaintain 0.388 -1.83 H YDogSwitch+ 0.386 -2.68 YMaintainDogcc = U.Ubx = number of pronominal references serving a specified reference function (switchor maintain) at T1.n = summed number of pronominal references serving a specified reference functionat T1 and T2.x/n = sample proportion (Pest)p= .5H1 : p< .561Table 5b. Comparison of pronominal reference to the boy and to the dog, for specifiedreference functions, T1 vs. T2 (joint reference to boy and dog excluded)x/n= p test statistic z Reject H0 Reject(z) if: H0?Switch 0.407 -1.97 -1.64 z<z YBoyMaintain 0.419 -1.27 NBoySwitch+ 0.411 -2.34 yMaintainBoySwitch 0.441 -0.686 NDogMaintain 0.340 -2.26 YDogSwitch+ 0.380 -2.18 YMaintainDogcx = U.Ux = number of pronominal references serving a specified reference function (switchor maintain) at T1.n = summed number of pronominal references serving a specified reference functionat T1 and T2.x/n = sample proportion (pea)H0 : p= .5H1 : p< .562of the dog (with respect to pronoun preference for switching reference) at both T1 andT2, consistent with Bamberg’s theory (tables 4a and 4b).Looking at the maintain reference function, there is no significant difference between T1and T2 in the preference for pronouns to maintain reference to each of the boy and thedog (see tables 5a and 5b). This is consistent with expectation and indicates that thechildren continued to choose pronouns to maintain reference almost exclusively. Indeed,in all cases but one (maintain reference, dog, T2), there were more examples of ellipsisas the form used to maintain reference than of full nominals (see tables 2a and 2b).Nonreferential Measures of Narrative CompetenceThree other measures were used to indicate the children’s ability to tell a “good story”at both T1 and T2. These were (1) the percentage of clauses contributing to thematicadvancement, (2) Mean Length of Clause (MLC) (where a clause refers to a clause whichcontributes to thematic advancement), and the “goodness” of the story based on thepresence/absence of six basic story elements (based on Berman, 1986). The results ofthese measures are shown below (tables 6 through 9).The percentage of thematic advancement clauses showed a wide range across all subjects(14% to 60% in the first telling, 46% to 79% in the second telling; see table 6). Fromtable 6 it can also be noted that for all subjects except Subject I the percentage ofthematic advancement clauses was higher on the second telling. When the data wassummed for all subjects, there was a significantly higher percentage of clauses63Table 6. Total number of clauses and number of thematic advancement clauses at eachtelling, individual and group; percentage of total clauses contributing to advancementof the story theme, individual and group, at T1 and T2Subject # clauses # thematic % thematicadvancement advancementclauses clausesT1 T2 T1 T2 T2 T2Si 73 97 44 48 60 49S2 45 34 18 20 40 59S3 69 141 30 78 43 55S4 82 46 43 35 52 76S5 56 42 20 27 36 64S6 69 29 10 23 14 79S7 62 37 17 17 27 46S8 66 51 29 27 44 53S9 60 44 13 33 22 75Sb 57 40 24 19 42 48All subj.’s 639 561 248 327 40 58Table 7. Comparison of the proportion of clauses which contribute to thematicadvancement out of the total number of clauses summed for all subjects, T1 vs. Tx1/n x2/n test statistic Zcrit reject H0 reject H0 ?(z) if:.388 .582 -6.74 1.64 Z<Zcritcc=0.05x # clauses contributing to advancement of theme summed for all subjects at aspecified timen = total # clauses, summed for all subjects at a specified timex/n= sample proportion for a specified time (pea)H0: P1 = P2H1: Pi <P264Table 8. Mean Length of Clause* for all subjects at T1 and T2Si S2 S3 54 S5 S6 S7 58 S9 SlO Ave.T1 4.i 3.8 4.9 5.7 6.2 5.4 6.2 5.2 4.4 4.9 5.08T2 4.4 6.7 5.4 5.8 4.7 6.2 5.2 5.7 5.4 4.4 5.39* Measured over clauses that contribute to thematic development.contributing to thematic advancement at T2 (see table 7). Further to this, except for twosubjects (Si, S3), the total number of clauses at T2 was lower than at T1. This means thatfor the most part the children were telling shorter, more focused stories at T2 (in thesense that more of their utterances were directed at the development of theme withinthe story).The mean length of clause (MLC) for each subject at each time is shown in table 8. Ascan be seen, there is variation among subjects. Seven subjects increased their MLC fromT1 to T2, and three decreased their MLC. The overall average for the MLC’s at T1 is 5.08,and at T2 is 5.39. This is not a large change, but the trend is toward longer clauses inthe second telling. An increase in MLC (or MLU) is often taken to be an indicator of anincrease in the complexity of constructions in a child’s language (e.g. Brown, 1973);therefore, it could be proposed that a longer MLC at the second telling is an indicatorthat the children are using more complex language in their second tellings.65As can be seen in table 8, Subject 2 showed the greatest increase in MLC. Examples ofclauses from her second telling which were responsible for the relatively large increasein MLC are as follows:There are two little frogs and a mom and a daddy down there.They’re going to be a sleeping boy’s.They gave him one of their little babies.The final nonreferential measure to be examined was story “goodness.” The presenceof basic story elements is a performance indicator of a child’s point in acquisition of astory schema. The six story elements considered basic to a “good” story are listed belowtable 9. The expected result was that children should include more of these elements atT2, after hearing an adult model repeatedly over the period of a week. This held truefor all but one subject (Si). Three subjects increased their number of basic storyelements by three, five subjects by two and one subject by one (see table 9). Thisindicates that the children were generally able to produce more complete stories afterbeing exposed to an adult model.It was not expected that the children would produce “perfect” stories after hearing theadult model, in light of the fact that children do not normally begin to produceorganized and complete stories (according to a story schema) until around age 5 (e.g.Botvin & Sutton-Smith, 1977; Applebee, 1978, both as cited in Liles, i993).66Table 9. “Goodness” of stories, as determined by presence/absence of six basic storyelements* (based on Berman, 1986), at T1 and T2T1 T2Story Elements Story Elements1 2 3 4 5 6 T 1 2 3 4 5 6 TSi ‘I ‘J ‘) .‘I X ‘15 I X X 1 J4S2 J X 1 “I X X3 ‘i X -J ‘15S3 q x ‘1 X X3 •‘1 ‘.1 ‘I ‘1 ‘J ‘16S4 ‘1 J J ‘J XX4 X ‘J ‘J J -15S5 1 X J ‘1 X X3 1 “1 1 1 ‘J X5S6 J X X J X X2 I x X X 4S7 1 X X X X2 q x 1 X X4S8 ‘Ix J ‘1 XX3 q ‘J X5S9 ‘Jxx ‘I XX2 1 ‘I -1 X5Sb x1 X ‘J X X2 ‘1 ‘1 ‘.1 -J X5*Story Schema Elements:1. Frog leaves jar2. Protagonist(s) discover frog’s absence3. Protagonists search for frog4. Boy finds frog5. Boy takes frog6. Frog is same as or substitutefor missing frog.The stories were still expected to reflect the findings of Orsolini (1990), who found that3- and 4-year-old children produce stories that contain incomplete and minimal episodes,even if a number of essential story elements are still present. The stories of the childrenin this study certainly were characterized by incomplete, minimal or absent episodes inboth conditions. The trend toward inclusion of more story elements indicates that the67children were able to incorporate more information in their stories to create a morecomplete story from the story schema perspective.The results of the three nonreferential measures show a trend toward children tellingmore developmentally advanced stories from various perspectives. They produce longerclauses, more efficient clauses (with respect to thematic advancement) and include moreelements deemed to be central to an adultlike story schema. This supports the proposedidea that with exposure to an adult model, children will have a greater opportunity toreveal in their productions what they know about stories. It also directly supports theconclusions of Merritt & Liles (1987), which indicated that children told longer, morecomplete stories in a story retelling task, although the basic structure of the narrativewas not significantly different from that in a generation task.However, as was described above, the data did not provide any indication that childrenwere more able to apply strategies for use of pronominal reference after their week ofreading the book with their caretakers. Therefore, Hypothesis 2 was not confirmed. Themost probable explanation for this is that any advantages resulting from exposure to theadult model in terms of allowing more opportunity for processing and producing theform/function pairing strategies for pronominal reference may not have been sufficientto result in significant differences in production or knowledge.68Comparison with Karmiloff-Smith (1981)The next two sections of the results present the data in ways which will allowcomparison with the results of Karmiloff-Smith (1981). The two main points of inquiryare (1) the extent to which pronouns are being used deictically, and (2) the degree towhich pronominal reference in clause-initial position is reserved for reference to the boy.The Use of Gesture (Point) with PronounsKarmiloff-Smith interpreted her data to indicate that children under the age of six wereusing pronouns deictically, not in a referential manner. After age six, children used athematic subject strategy (preferential use of pronoun in utterance-initial position to referto a protagonist). Her main argument to support her conclusion was that mostutterances that contained pronouns were isolated statements, static picture descriptionswith no apparent intention on the part of the child to link them linguistically. Thisstatement was entirely observational and not supported by specific evidence from herdata or that of other researchers. The only data-driven evidence she gives was at thesingle utterance level, and this was: (1) the high frequency of spatial deictics and (2) thehigh frequency of pointing that accompanied pronoun use. Because most subjects in thepresent study were videotaped, it was possible to examine the data collected here withregard to point.There is no doubt that in the 3-year-old children in the present study used a substantialamount of deictic language. In the stories there are such unmistakable examples ofdeictic nonnarrative language as description of pictured background information,69occasional use of the indefinite article “a” to reintroduce an established character, and(relatively frequent) use of constructions like There’s a X, I see a(n) X, or This/That/TheseX. In fact, for the most part the stories were similar at a glance to the short example ofa 4-year-old’s story provided by Karmiloff-Smith; there was frequent use of theconnectives and then at the beginning of clauses, and the clauses often had the feelingof being “isolated” as Karmiloff-Smith described them. It is dangerous, however, to relysimply on “feelings,” and a measure was sought which would allow empiricalassessment of the nature of pronoun use. Therefore I set out to determine whetherpronouns were being used in a referential rather than only deictic manner, despite thesignificant presence of other deictic constructions. The measure which seemed mostappropriate and rigorous for analysis was use of point in accompaniment of pronouns,an indicator of deixis Karmiloff-Smith noted and one which has been noted by othersas well (e.g. Wales, 1986; Tomasello et aL, 1984/85).Simply counting the number of pronouns accompanied by point is not an informativemeasure, due to the nature of the experimental task. The book served as a joint focusof reference between the children and the experimenter, and the experimenter did notengage in any talk which would normally accompany a joint story “reading” (asdescribed in chapter 2). Therefore, the task generally promoted the accompaniment ofutterances with point, as a way for the children to request feedback and interaction,which were unnaturally absent during their story tellings. From this, the clausesaccompanied by point were not expected to be exclusively deictic in construction andfunction. Therefore, it was deemed to be more informative if the number of points70which accompanied a pronoun (in clauses important to thematic advancement of thestory, i.e. coded clauses) were compared with the number of points which accompanieda full nominal (which by definition is not deictic). If there are no significant differencesbetween the number of points that accompanied pronominals and the number of pointsthat accompanied nominals, then it cannot be concluded that the point whichaccompanies a pronoun is a strong indicator of deictic use.These results were then compared with results showing the difference between thenumber of deictic terms accompanied by point and the number of deictic termsunaccompanied by point. It was expected that there would be significantly more deicticterms that were accompanied by point than those which weren’t.Important to remember during the discussion of these results is that the numbers fornominals and pronominals in tables lOa and lOb include only those forms that occurredwith a point. Therefore, these numbers are only a portion of the total numberspresented above in tables 2a and 2b.Tables lOa and lOb contain summed “point data” for all subjects, and are similar exceptfor the exclusion of the counts for Subject 3 in table lOb. Subject 3 seemed to beapplying a different strategy in her story telling with respect to pointing. She pointedwith almost every utterance regardless of form or function, especially in her secondtelling. This indiscriminate pointing was not seen to the same degree in any othersubject. Furthermore, she had by far the longest stories of all subjects, and therefore a71Table lOa. Gestural data summed for all videotaped subjects*: Number of pronominaland nominal full referential forms and deictic terms with point, number of deictic termswithout point, and number of occurrences of point with terms other than deictic or fullreferential forms, at T1 and T2Category Number of Occurrencespoint+full pro 18referential nom 13termpoint+other 15T Deictic Term:1 these see/I see there’s!this/that look (right) therepoint + deictic term 42 7 25no point + deictic term 15 3 8point+full pro 57referential nom 23termpoint+other 38T Deictic Term:2 these see/I see there’s!this/that look (right) therepoint + deictic term 41 4 25no point + deictic term 4 10* Subjects 2 and 10 were not videotaped and therefore are not included.72Table lOb. Gestural data summed for all videotaped subjects*: Number of pronominaland nominal full referential forms and deictic terms with point, number of deictic termswithout point, and number of occurrences of point with terms other than deictic or fullreferential forms, at T1 and T2 (excluding subject 3)Category Number of Occurrencespoint+full pro 6referential nom 11termpoint+other 9T Deictic Term:1 these see/I see there’s/this/that look (right) therepoint + deictic term 2 6 12no point + deictic term 3 8point+full pro 16referential nom 16termpoint+other 25T Deictic Term:2 these see/I see there’s/this/that look (right) therepoint + deictic term 27 4 19no point + deictic term 3 10* Subjects 2 and 10 were not videotaped and therefore are not included.73Table 11. Comparison between summed number of deictic terms accompanied by pointand number of deictic terms unaccompanied by point, at T1 and T2, with and withoutSubject 3x/n test statistic Zcrit reject H0 reject H0?(z) if:T1 .580 .898 1.64 z>z Nwith S3T2with S3 .713 3.80 “ I’T1without S3 .353 -1.21 I? ‘I NT2without S3 .500 0 NU = U.U3x = number of pronominal references accompanied by point at a given timen = summed number of pronominal and nominal references accompanied by pointat a given timex/n = sample proportion (p)H0:p=.5H1 : p> .5difference in pointing strategy on her part would strongly affect the group data. It wasof interest to see whether and how the results would differ with her excluded from thedata pooi.The results indicate that when Subject 3 was included in the data pooi, at T1 there wasno significant difference in use of point for nominals and pronominals, but there wasno evidence at T2 that point was evenly distributed between nominals and pronominals,and the null hypothesis to that effect was rejected in favour of there being more points74attached to pronouns (see table 11). However, as predicted above, when Subject 3 waseliminated from the data pooi, there was no significant difference in the distribution ofpoint with pronouns and full nominals at either telling. In contrast, it was found thatat both times, with or without Subject 3, there were significantly more deictic termsaccompanied by point than those unaccompanied by point.These results support the expectation that pronouns are not being treated in the sameway as a deictic term (with respect to one indicator of deixis). Furthermore, the resultsshow no difference in the treatment of pronominals and nominals. Therefore, theseresults dispute Karmiloff-Smith’s conclusion that all children under six years are usingpronouns in an exclusively deictic manner. The results do not confirm Hypothesis 3a.These results also disagree at first look with the findings of Tomasello et al. (1984/85),who found that children between 1;8 and 3;7 pointed with pronouns more frequentlythan with nouns. However, caution must be exercised in comparing the results of thepresent study with those of Tomasello et al.. Their study was not a narrative task, buta series of elicitation tasks which included such elicitation techniques as feigned adultnoncomprehension of a child’s request. The difference between the two outcomes couldbe taken as support for the proposal central to this study: that the pronouns foundwithin the context of the narrative task are serving functions which are constrained andguided by “global aspects of the narrative, i.e. creation of coherence within the text.75This would be in keeping with the results of Bennett-Kastor (1983), who showed thateven 2-year-olds are capable of certain linguistic aspects of coherence in text, such assensitivity to new and given information by the use of definite and indefinite articles aswell as the use of pronouns to reiterate characters within a narrative task. They werealso very sensitive to the function of the grammatical subject to create focus around acharacter, an aspect of Karmiloff-Smith’s thematic subject strategy to be discussed in thefollowing section. Despite this linguistic capability, the 2-year-old children were notadvanced enough in other aspects of narrative production (e.g. cognition and semantics)for these linguistic capabilities to function as strategies for creating global coherence suchas thematic advancement. Bennett-Kastor found that by five years of age, children weregood at using linguistic capabilities to serve global cohesive functions, and were not sotied to the syntactic subject position for mention of a character in focus. The results ofthis study with 3-year-olds would place their performance somewhere between that ofthe 2- and 5-year-olds in Bennett-Kastor’s study. They showed referential use ofpronouns which was sensitive to thematic advancement, but did not tell stories ascomplete as the 5-year-olds with respect to a story schema. Regardless of exactly wherethey fit into the developmental time-line of Bennett-Kastor’s results, both studies foundevidence of pronoun use constrained by global elements of story cohesion before the ageof six.Distribution of Referents to Pronouns in Initial PositionKarmiloff-Smith and Bamberg have in common the proposal of an early thematicadvancement strategy which reserves the use of pronouns for the protagonist in76switching reference, although they disagree on the age at which this strategy appears.Karmiloff-Smith adds another element to this strategy. She proposes that, in addition,when a pronoun is initial position, it is reserved for reference to the protagonist, and sheprovides a number of examples of stories in her study for which this holds true.In the present study it was found that even when the pronominal references in initialposition to the boy alone are summed with the joint pronominal references to the boyand dog, these only constitute 105/204 (51%) of all pronouns in initial position (see table12). This certainly does not point to a strong preference to reserve initial pronouns forreference to the boy (protagonist). The same calculation for switch reference functionalone shows a higher degree of preference for the boy (65% of the initial switch referencepronouns are to the boy, (including joint reference to the boy and dog)). Hypothesis 3bis therefore not confirmed.At first glance, there does not seem to be anything particularly special about the clause-initial position for reference to the boy. However, an interesting result was found whenclauses with initial pronouns serving the maintain reference function were examined.A new category was devised for this particular look at maintain reference, calledmaintain reference (local). This referred to clauses which contained a pronoun (in thesecases clause-initial) which referred to a character previously established or re-establishedby a full nominal phrase. Either the full nominal occurred either immediately prior tothe pronominal reference in question, or intervening utterances contained reference only77to the same character (i.e. no other character was mentioned in these interveningclauses).Of the clauses with initial pronouns serving to maintain reference, 36% were “local,’ andwithin these 36% there were absolutely no examples of maintain reference (local) to theboy (or boy plus dog) (see table 12). A low number of examples of maintain reference(local) was predicted for the boy due to the limited number of full nominals used torefer to the boy at any time for any function. However, the fact that there were noexamples of this for the boy seems noteworthy. This can be interpreted not so much asa comment on how children use pronouns to maintain reference (it is expected that theywould lean toward use of pronouns for the maintain reference function regardless ofreferent), but more as a comment on the pattern of introduction and maintenance ofreference to characters other than the protagonist when they are brought into thenarrative during various episodes. This is just another way in which children appearto be treating the protagonist and other characters differently with respect to pronominalanaphoric reference.Although the data do not seem to support Karmiloff-Smith’s notion of reservingutterance initial occurrence of pronouns for reference to the protagonist, this differencemust be discussed in the context of differences between the design of the two studies.As noted in the introduction, the length of the books (in terms of pictures), the numberof episodes and the number of central and less central characters are much greater inthis (and Bamberg’s) study than in Karmiloff-Smith’s. To put these differences into78Table 12. Utterance-initial pronouns used by each subject at T2 to refer to the boy andjointly to the boy and dog, shown as total and by switch reference, maintain referenceand maintain reference (local)* functions# total total# #SR #SR #SR #MR #MR #MRS Initial # to to B+D to to (L) (L)Pron’s boy boy B+D (boy)1 36 13 -- 11 6 -- 25 15 --2 10 1 2 4 1 2 6 5 --3 65 22 7 36 13 7 29 5 --4 11 7 1 7 4 1 4 1 --5 15 11-- 9 8-- 6 3 --6 11 3 5 5 2 2 6 3 --7 9 1 6 4 1 2 5 -- --8 16 6 2 10 4 2 6 2 --9 17 8 1 12 7 1 5 2 --10 14 7 2 6 3 2 8 -- --All 204 79 26 104 49 19 100 36 --* Maintain reference (local) indicates utterances which contained a pronoun to referto a character previously established by a full nominal. Either the full nominaloccurred immediately prior to the pronominal reference, or a small number ofintervening utterances occurred which contained reference (pronominal) to thesame character only.perspective, the entirety of Karmiloff-Smith’s story could be equated to one episode inthe search motif in Frog, Where Are You?. It is possible that the children in KarmiloffSmith’s study would have altered their approach to reserving initial position forreference to the protagonist with the increased complexity and length of the story, andthe increased number of characters. (Or, of course, the possibility remains that they arenot employing this strategy.) Furthermore, it is difficult to compare my data directlywith the data from Karmiloff-Smith’s study because she did not include any numbersrepresenting the percentage of initial pronouns which referred to the protagonist. It is79impossible to say whether the obtained result that 65% of initial pronouns used to switchreference had the boy as referent is comparable to her results.To look in a different way at this notion of clause-initial position being important whenmaking reference to the protagonist, the percentage of total pronominal references to theboy which were in clause-initial position was calculated for the switch reference functiononly, as well as for the summed switch and maintain reference functions. These werecalculated to see whether the clause-initial position was preferred for reference to theboy. For the switch reference function alone, 56% of the pronouns were in initialposition. For both reference functions summed, 86% of the pronouns were in clause-initial position. This indicates that initial position is preferred when referring to the boy,especially for maintaining reference. This can be explained in the following manner.Although formal counts weren’t made, in most clauses (especially those which containeda maintain reference) the protagonist was the subject of the clause, and thereforereference to the protagonist was found in initial position. In contrast, a higherproportion of clauses containing a switch reference to the protagonist had theprotagonist (boy) as the object of the clause, naturally found in a noninitial positionwithin the clause (e.g. “Then the reindeer throws jj down into the river”).As Bennett-Kastor (1983) noted in her study of noun phrase coherence in narratives, thecharacter which is chosen to be focused upon in all or part of the narrative is usually theagent of clauses in which it is mentioned, or at least the grammatical subject. It wouldfollow that the utterance-initial position would be nearly exclusive for mention of the80protagonist in a story where focus never moves away from that character (as inKarmiloff-Smith’s balloon story). However, because the protagonist of Frog, Where AreYou? is not the main character of all episodes, focus changes from character to character,and in several episodes the protagonist is the object, not the agent, of actions (as in theexample above). Therefore, it is proposed that due to the story’s greater length andcomplexity, in this study more grammatical roles were used for mention of theprotagonist, in particular for switching reference. This is supported by Bennett-Kastor(1983), who found that by five years of age (when children are telling longer, morecomplete stories in a generation task with no visual supports) there is less tendency tomention characters in focus only as grammatical subjects.The result above indicates that Karmiloff-Smith’s decision to only look at pronominalreference in initial position limited her ability to assess use of pronouns to makereference within stories, and also limited her scope with respect to the role thematicconstraints play in the use of anaphoric reference. Although it was not a goal of thisstudy to closely examine the distribution of form/function pairings with respect tosubject and object positions within a clause, this should not be ignored when assessingthe way in which children organize the clauses within their stories to create a cohesivenarrative. Certainly, Karmiloff-Smith missed a significant percentage of the totalpronominal (and nominal) references made to characters within the stories by excludingnoninitial occurrences. Indeed, these are points she raised herself when discussing thescope of her study.81Across-Subject MeasuresBecause theories of the development of aduitlike form/function pairing for anaphoricreference (like those of Bamberg and Karmiloff-Smith) are based on the notion of adevelopmental time-line, it was of interest to see whether and how different indicatorsof development (age, mean length of clause, and “goodness” of story) would correlateacross subjects. Further, it was desirable to see whether the data would “split” on ameasure of pronominal reference development for the younger and older subjects.As can be seen in table 13, there was no significant correlation between any of thedevelopmental measures listed above. It was not strongly expected that chronologicalage would correlate with the other two measures, due to the widely accepted normalvariability in the developmental rate at which children acquire language (the reason theMLU measure was created in the first place; Brown, 1973). In this group of subjects,with a range in age from 3;2.3 to 3;9.12, a wide range in developmental levels would beexpected, not necessarily tied closely to age. It was more expected that MLC and“goodness” of story would correlate, because each measure is independent ofchronological constraints. This correlation was not found. However, MLC is often takenas an indicator of children’s developmental language level, particularly the complexityof syntactic constructions. There is no reason to assume that their development of storyschema should be directly tied to (i.e. correlated with) development of languageelements, even though they are developing in parallel and do interact (for example achild who does not have a full range of pronouns could not have a highly developedsystem of pronominal reference in narratives).82Table 13. Test for positive linear correlation between across subject measures: MeanLength of Clause (MLC), Goodness of Story, Age (months)T y x test statistic Reject H0 Reject(r) if: H0?T1 MLC age 0.0165 0.549 r>r N(months)T2 MLC age 0.115 N(months)T1 MLC goodness -0.272 U NT2 MLC goodness -0.0461 NT1 good- age 0.429 Nness (months)T2 good- age 0.106 Nness (months)cz = 0.05no significant positive linear correlationH1 : significant positive linear correlationr = correlation coefficient (r=1 : perfect linear correlation, r=0 : no linear correlation)goodness: “goodnesst’of story as measured by presence or absence of six basic storyelements (See table 9: Berman, 1986)83Table 14. Comparison of proportion of pronominal reference out of total reference madeto the boy serving the switch reference function between the oldest four subjects and theyoungest four subjects*, summed for T1 and T2 (joint reference to boy and dog included)x1/n x2/n test statistic z1 reject H0 reject H0?(z) if:0.803 0.907 -1.72 4.64 Z<Zcrita U.05x1 = summed number of pronominal reference to the boy serving switch referencefunction, four youngest subjects, T1 + T2x2 = summed number of pronominal reference to the boy serving switch referencefunction, four oldest subjects, T1 + T2n1 = summed number of reference (all types) to the boy serving switch referencefunction, four youngest subjects, T1 + T2x2 = summed number of reference (all types) to the boy serving switch referencefunction, four oldest subjects, T1 + T2x/n = population proportion (Pest)H0: pi = P2H1 : Pi <P2*Note: Two subjects (S4, S6) had identical ages. Their age fell at the mid point, withfour subjects higher in age and four subjects lower. Their data were not used for thismeasure.Interestingly, however, the proportion of pronouns used to switch reference to the boywas not equal for the young half of the subjects and the old half, and pointed towardthe younger subjects using fewer pronouns to fill that function (see table 14).These results are consistent with the expectation that children of this age range wouldbe in the process of moving away from deictic use of pronouns, and moving towardacquisition of the first proposed stage in the development of pronominal anaphoricreference, namely the favouring of pronouns to switch reference to the protagonist. This84is consistent, too, with the fact that while the narratives of these children containedfrequent examples of deictic constructions and often retained the feeling of isolatedpicture descriptions, there was a significant difference between how the children treateddeictic constructions and pronouns with respect to gesture, as well as significantevidence to support Bamberg’s early thematic advancement strategy. Further, as willbe discussed below, errors the children made and corrected also point towarddevelopment of an anaphoric reference strategy guided by attention to thematicadvancement.Errors and Self-CorrectionsThis study also contradicts Karmiloff-Smith’s conclusion that children under th& age ofsix demonstrate no intersentential linking related to thematic advancement, partly dueto general compatibility with the results of Bamberg’s study, but more specifically dueto occurrences of ?errorsH (perceived by the child), which were often followed byspontaneous self-correction. A number of these errors occurred when children werechoosing a pronominal or nominal form for a particular reference function,demonstrating a purposeful use of the thematic subject strategy. Other errors and selfcorrections were those which reflected nondeictic use of pronouns, as well as thosewhich revealed knowledge of and sensitivity to elements of local cohesion and thematicadvancement.The errors which occurred when choosing the appropriate form to make referenceseemed to indicate an apparent struggle on the part of an individual child to choose the85appropriate reference form (nominal or pronominal) to switch reference to the dog.These errors reflect the constraints of thematic advancement on the developing use ofpronominal reference and support the proposed preferential use of pronouns to switchreference to the protagonist, and use of nominals to switch reference to other characters.The error examples are shown below. All are examples of clauses in which the functionof the reference is to switch reference to the dog.1. (He p>) The doggy put (the head) him’s head in the jar.2. And (he f>) (the doggy) (the) the bees chased the doggy.3. And then (it) the doggy looked in the jar.4. (He) the dog went outside.5. (And then he 1>) And then the dog’s put it on his head.6. (He was looking) the dog was looking up. And he waslooking up. (switch reference to boy)Example 1 shows obvious self-correction from use of a pronoun to switch reference tothe dog to use of a full nominal. Interesting in this example is the apparentperseveration of this “rule,” whereby the head is used to talk about the dog’s own head,but is subject to on-line correction to a possessive pronoun (him’s head). Examples 3through 5 show similar self-corrections to change the pronoun to a full nominal inswitching reference to the dog. In example 2, the he in the sentence fragment he f> wasactually interpreted as referring back to the boy who had been referred to in theprevious utterance. However, the juxtaposition of the maintaining he in reference to the86boy and the switching the doggy in reference to the dog shows the contrast of treatmentfor the two characters. The subsequent self-corrections seem to be a result of the child’sattempt to avoid a passive construction.There are other examples of self-corrections that gave insight into the developmentalprocesses some of these children brought to the story-telling task, although they may notbe classified as “errors.” One self-correction in particular supported the proposal abovethat, although there was a significant number of deictic constructions in the stories, andalthough many strings of utterances appeared to be isolated picture descriptions, therewas a developing nondeictic treatment of pronouns. This is supported by the followingstring of clauses:7. Then he’s dropped in water.(Boy, SR)(he accompanied by point)And then he sinks. (Boy, MR)Researcher: Pardon?(Then he) This guy sinks. (This guy accompanied by point)The most important clause in the string is the last clause. This clause follows theresearcher’s request for repetition of information already given. The child begins byusing a pronoun to refer to the character again (no point), but then switches to thedeictic construction this guy accompanied by a point to the page. This purposeful switchto a deictic construction in order to ensure joint attention of the subject and researcherto the character in question implies that the initial use of the pronoun was not deemedby the subject to be capable of establishing that joint attention. Furthermore, the87pronoun was not accompanied by a point, which also indicates a move toward nondeictic use of the pronoun. One could argue that this is an example of one deictic termbeing replaced by a “stronger” deictic term, but in the context of other informationsupporting nondeictic, referential use of pronouns this is not likely the case.Another self-correction led to the observation that some of the children sometimesseemed to be attempting to use it in reference to the frog and/or the dog as a way todistinguish them from the boy and/or from each other. In the context of this storyeither he or it would be appropriate to use in reference to the boy and the frog, but someof the children seemed to be struggling in making the decision of which one to use. Theself-correction which instigated this observation was as follows:8. (Subject 5)When the frog creeped out.1k creeped out.And then he said, “Woof!” that he gone. (1.dog, 2.frog)Then (j) g got out the window. (frog)The child in this case (subject 5) was about to use it to refer to the frog in the final clausewhen he apparently realized that he had already used he to refer to frog in earlierclauses and self-corrected. The child did not seem sure about which pronoun would bemost appropriate to use in reference to the frog. He seemed to be his most instinctivechoice (and this is true for all subjects) but it is seemingly brought forward when the88possibility exists that the listener might confuse referents. On its own, this self-correction did not seem significant, but in other stories, this apparent strategy to use itas a further distinguisher of referent (although it was not consistently attempted) wasrepeated, as the following examples show:9. (And the) and the frog is (in his)(in his) in his trap jar.li popped out the window. (frog)Then was gone.And they’re looking all over for . (1.boy+dog, 2.frog)10. j broked the glass jar. (SR, dog)He broked it. (MR, dog)In example 9, the child chooses a combination of pronouns chosen to make reference tothe frog. In this case, too, it seems as though there is an incompletely formed rule forthe child:use it when necessary to further specify a referent of a pronoun.This also seems to be the case in example 10, where the child chooses to use it to switchreference to the dog, but reverts to he to maintain reference. There were not enoughexamples of this use of it to refer to the dog and/or frog to make generalizations abouta strategy which uses the inanimate pronoun to make unique reference to each character,but these examples provide enough information to propose that at least some of thechildren in this study were attempting to employ such a strategy.89Two other self-corrections (both from Subject 5’s second narrative) indicate activeattention to theme and development of a story-line. These were as follows:11. And then (the) a squirrel> (first mention of squirrel)12. They looked at their frog to sleep.“I wanna go out and see my family.”The frog said, “I’m gonna go out and visit my children.”(And then the f>)Then they went and looked everywhere and everywhere and everywhere andeverywhere and everywhere and everywhere and everywhere and everywhere,everywhere, everywhere.(turning pages with episodes from the search motif)Then they found frogs.Then he brang them back home.(third version of the story, T2.)In example lithe subject corrected the use of the definite article the to the indefinitearticle a for first mention of a character, an indication that he was aware of the discourseproperties of these two articles. But example 12 is more illuminating with respect to thesubject’s understanding of theme. He had already told two versions of the story at T2and this was his third. He began the story, but after a few clauses, instead of continuingin detail about all episodes in the search motif, he replaced them with they went and90looked everywhere and everywhere. ..(etc.). This is clear evidence that at least this subjectwas not treating the task as a series of isolated picture descriptions.Interestingly enough, Subject 5 was the oldest subject in the group, and the only subjectto condense the story in the manner shown above (but also the only subject to attemptthree versions of the story at one telling). However, he was only 3;9.12 at the time ofhis first telling, so he provided evidence of attention to theme in his stories at an agemuch younger than that recognized by Karmiloff-Smith. The results of developmentalmeasures (other than pronominal reference) for Subject 5 were not particularly advancedwithin the group as a whole. He did not have the highest MLC averaged over the twotellings (although it was the highest at T1), he did not show the highest number of storyelements at either time, and although the number of deictic constructions in his storieswas the lowest, it was not remarkably different from the rest of the group as a whole(it was only one lower than for Subject 6). His use of pronouns to make reference wasa near perfect fit for Bamberg’s model; all but one switch reference to the boy waspronominal, and all but one switch reference to the dog was nominal (and that referencewas marked “appropriate”).The point here is that while Subject 5 showed more direct evidence of attention to themethan other members of the group, he was not the most developmentally advanced onall measures. Therefore, it is proposed that the sensitivity to elements of narrativestructure he demonstrated in the above examples is probably not unique to him. It mayalso be possible that aspects of story structure are independent of each other, as was91shown for MLC and “goodness” above. Although other subjects did not reveal theirknowledge so explicitly, they most likely possessed the same knowledge and sensitivityor were in the active process of acquiring such knowledge.Individual DataBamberg noted that although the tendency of the group was to use pronouns to makereference in a bifunctional way, namely to switch reference to the main protagonist (butnot other characters) and to maintain reference to all characters, there was still apossibility that individual children were using different strategies to make reference. Henoted that in the group data, the number of nominals used to switch reference to the boyand the number of pronominals used to switch reference to the dog were greater thanzero. He therefore went back to the stories of individual children and tried todifferentiate the strategies they use to make reference. In particular, he recognized thepossibility that all children may not follow the same developmental path to reach theadultlike anaphoric use of pronouns and, therefore, “might follow different strategies ofthematically working their way through the story” (1987: 79).In this study, too, the general pattern of pronoun use supported that found by Bamberg,but the number of nominals used to switch reference to the boy and, in particular, thenumber of pronouns used to switch reference to the dog brought the individualstrategies of the children into question. The following section takes a closer look at theresults of nominal-pronominal distribution in individual stories and the contexts withinwhich these forms were used in ways that go against expectation.92Use of Nominals to Switch Reference to the BoySeven out of ten subjects told stories that had examples of switching reference to the boywith a full nominal expression. Across all subjects and both tellings, there were fourteenexamples in all. When these examples were examined for their linguistic constructionand the context in which they were used, most fit into one of four different categories:(1) deictic constructions, (2) proper nouns, (3) story conventions and (4) question forms.Subjects 3, 4 and 7 demonstrated the most frequent use of nominals to switch referenceto the boy (three examples each). For many of these, the subjects used truly deicticnominal expressions to serve the switch reference function (that boy, this guy, this boy).These deictic constructions were all accompanied by point. Subject 4 (3;6.2) used suchconstructions frequently throughout her narratives (with respect to all characters), andthe number of deictic constructions accompanied by point which referred to the boy anddog was higher for her than for all other subjects. This frequent use of deicticconstructions accompanied by point to draw attention to different characters, incombination with the fact that switch reference to the dog was almost evenly distributedbetween nominal and pronominal forms, seems to indicate a system of reference that isstill developing, moving from the deictic use of expressions to distinguish characterstoward nondeictic use (influenced by the interconnective aspects of narratives such asthematic advancement). However, there is no indication that Subject 4 is developing astrategy of pronominal reference different from that of the majority of children who willcome to use the thematic advancement strategy, especially in light of the fact that shenever used pronouns to switch reference to the dog alone, only jointly to the boy and93dog (a point that will be discussed in the next section). She simply seemed to be at anearlier point in development of the first stage of acquisition of pronominal anaphoricreference.Subject 3 did not use deictic constructions to refer to the boy to the same extent asSubject 4, but showed immaturity in her use of pronouns for making reference in otherways. She accompanied almost all of her clauses with a point, and in particular her useof point with pronouns was more frequent than for any other subject. This is anindication that she too was still in the process of abandoning purely deictic use ofpronouns in the development of the thematic advancement strategy. At first glance sheseemed more advanced than Subject 4, because she favoured pronouns more often thatSubject 4 did for switching reference to the boy. However, when it was observed thatshe also almost exclusively used pronouns to switch reference to the dog, the presenceof even an emergent thematic advancement strategy in her narratives was questioned.There is no difference in how she uses pronouns to refer to the boy and the dog; shefavours the use of pronouns to switch reference to both characters. Subject 3 may be theonly example in this data set of a child who is not yet using pronouns in a non-deicticfashion.In two instances, a deictic full nominal switch reference to the boy was the first mentionof the boy in the T1 story: Subject 7’s only deictic construction and one from Subject 4.While the low number of examples does not allow more than speculation, it may be thatat the beginning of the narrative, especially when the child is unfamiliar with the story,94the protagonist is not yet chosen and strategies for thematic advancement are not yet inplace. This possibility of the protagonist not being a preselected, unchanging characterwill be revisited in discussion of reference to the dog below.Proper nouns were the second interesting use of nominal expressions for switchreference to the dog. Examples of this were found only in the T2 stories for Subjects 3and 7. There is little doubt that the names for characters were introduced to the childrenby their parents’ narratives. Indeed, the children had a difficult time remembering thenames which had been assigned to each character, and in both the stories, attempts tokeep referring to the boy by name were abandoned for most of the story. Both Subjects3 and 7 only switched reference to the boy by name once, in the opening episodes of thestory. Therefore, this use of proper nouns, while technically a nominal expression, doesnot seem to strongly violate the strategy of using nominals to switch reference to theboy.The third category into which a number of nominals fit was “story convention.” Forexample, the only nominal expression used by Subject 1 to switch reference to the boyis in the very first introduction of the character in her T2 telling: There was a little boy.This is the opening clause of her narrative, preceded only by a title: Froggy where are you.This is an example of convention in story-telling, probably a direct result of the way inwhich her caregiver(s) told her the story during the week. This is an entirelyappropriate use of a nominal expression for switching reference to the boy, not onlyreflecting adult reference strategies, but also reflecting story-telling conventions. The95subject’s manner while saying the title and this first clause seemed to be an imitation ofher caregiver and was soon abandoned in favour of her natural speaking manner. Fromthat point on, her story contained no more examples of nominals for switching referenceto the boy (and only one example of a pronoun used to switch reference to the dog).In other words, this story contained strong evidence for use of the thematic advancementstrategy. The conventional beginning to the story was complimented by a conventionalending: He take one of the baby. He tooked him home. That’s the end. This appropriateattention to story-telling convention may provide insight into the child’s developingunderstanding of theme and narrative structure, but it certainly does not detract fromthe strong application of the thematic advancement strategy throughout her story.Question forms were the fourth category into which examples of nominals used toswitch reference to the boy. In two cases, (found in the stories of Subjects 1 and 7), theclauses which contained such nominals were in the form of questions. Most questionsdirected to the researcher were coded as narrator comments and not included in thecounts for reference. However, if a question was used to divulge information importantto development of the story (and that information was not repeated in any other clause)then it was coded. For Subjects I and 7 this was apparently a style for creating theirnarratives. For both of them, questions were frequent and a number were coded (e.g.Why did that deer push the boy down?). In this clause, the fact that the deer pushed theboy down is important information for that episode, so the utterance was coded).96However, although the decision to include clauses of this type was made in order togive credit to the children for their knowledge of story content, it created aninconsistency with other clauses included in the coding. By asking these questions, thesubjects “stepped out” of the narrative, and directed these utterances at the researcher.Although the questions contained important story information, they were not equivalentin grammatical form or in speech act type to other coded clauses. The way in whichchildren referred to the boy during these questions therefore should not be expected toreflect strategies used in the narrative itself. In this sense, the fact that the children useda full nominal to refer to the boy in these question clauses may have no direct relationto the way they refer to the boy in clauses truly within the narrative.This leaves only three examples of nominal forms used to switch reference to the boyunaccounted for (out of 24 nominal switch references to the boy). These can be viewedas true violations of the thematic advancement strategy. The contexts in which theseexamples occur do not provide any certain clues as to the processes by which thechildren chose to use the nominal at that specific point in their narrative.Use of Pronominals to Switch Reference to the DogSimilar to the preceding section, there are several points of discussion which can explainmost of the cases in which subjects used pronouns to switch reference to the dog,apparently violating the thematic advancement strategy, which predicts exclusive use ofnominals to serve that reference function.97First, half of the examples of pronouns used to switch reference to the dog were actuallyjoint references to the boy and the dog. There was only one example in all the data ofthe use of a full nominal to switch reference to the boy and the dog. As discussedearlier, it is proposed that when the reference is joint to these two characters, the factthat the boy (the protagonist) is part of the referent is the overriding influence and thethematic advancement strategy rule that requires a pronoun to switch reference to theprotagonist is followed. Bamberg does not cite any occurrences of joint reference in hisdata, and does not indicate whether or how he coded joint reference if it did occur.However, by keeping the joint references separate in coding the clauses in this study,this “override” principle became clear.The second main group of clauses in which pronouns switched reference to the dogwere clauses describing episodes in which the dog was the main character, in particularthe episodes in which the dog breaks the jar and the scene in which the dog is involvedwith the bees. The fact that the dog is the “protagonist” of these specific episodes mayaccount for the higher occurrence of pronouns used to switch reference to the dog. Thechildren may be tending to apply the rule:use pronouns to switch reference to the protagonistto the dog in these episodes. The occurrence of pronouns used to switch reference tothe dog in episodes where the boy and dog are both involved is relatively lower.Further, during these same episodes at least three children used the inanimate pronounit to fill this switch reference function, even though they used “he” to refer to dog98elsewhere in the story. For example, Subject 2 use the following clauses to talk aboutthe “jar scene”:13. Oops!broked the glass jar. (SR, dog)He broked it. (MR, dog)In the above example, it is used to introduce the dog, but there is an immediate returnto he as the pronoun of choice in referring to the dog. Again, while there are fewexamples of this shift in pronoun choice from he to it in cases where a pronoun is usedto switch reference to the dog, it seems to indicate the child’s need to keep the dogseparate from the boy for the reference functions, and perhaps a lack of “comfort” whilechoosing pronouns to switch reference to the dog.Third, several other occurrences of pronouns serving to switch reference to the dog werethose deemed “appropriate” (SR) in the coding category. The definition of the“appropriate” category was outlined in chapter 2, but to reiterate, the context for an“appropriate” use of pronoun to switch reference to a character other than the boy wouldbe similar to the following:14. trying to chase that dog. (deer,SR; dog,SR)g trying to chase the dog. (deer,SRa; dog,SR)9915. looked in the jar. (dog,MR)And then doesn’t see (dog,MR; frog,SR)And then failed out the window (dog,SRa)Such contexts for use of the pronoun were often restatements of the content in precedingclauses (e.g. 14 above). Because the second clause contains the same subject and objectin identical sentence positions, and repeats the information in the first clause, there is nodoubt as to the referent of the pronoun in the second clause. These clauses, althoughthey could be considered to not move the story-line along because they repeatedpreviously stated information, were counted; they often added a small element of newinformation. Other examples were not restatements, but still maintained referents insubject and/or object position such that no ambiguity was possible (e.g. 15 above).However, as was the case in the preceding section, the decision to include these clausesin the coding created an inconsistency which resulted in apparent violation of thethematic advancement strategy. Technically this is true, but the clarity of the context(where all opportunities of ambiguity are removed by maintaining sentence positions forreferents) seems to be sufficient to allow the children to apply local rules of cohesion.The narrow range of contexts in which this occurred seems to be an exception to, ratherthan a violation of, the more “global” strategy.Just as there were a few examples of full nominals used to switch reference to the boythat could not be interpreted as anything but violations of the thematic advancement100strategy, so there were a few such examples of pronouns used to switch reference to thedog. This is not at all troubling when one considers that in the narratives of thesechildren, a developing system of reference is being observed. It is a positive point inthis data set that examples of deictic use of language, and “mistakes” in referencestrategies were made. This gives confidence to the interpretation that these childrenwere “caught” in the process of developing this proposed first stage in the acquisitionof true anaphoric pronominal reference.101CHAPTER FOURSUMMARY AND CONCLUSIONSGeneral SummaryThe general purpose of this study was to explore the way in which 3-year-old childrenuse pronouns to refer to characters in a narrative. The use of pronouns was examinedfrom a functionalist perspective, looking at not only the pronoun forms used, but thecohesive reference functions those forms served throughout the narrative to createcoherence for the text as a whole.There were three main goals of this study. First, the stories underwent analysis todiscover if they showed use of the thematic advancement strategy, as would bepredicted from the theory of pronominal reference development proposed by Bamberg(1987). Second, the performance of the children on measures of pronominal referencewas compared for the conditions in which children were unfamiliar and familiar withthe story, in order to detect any change in performance as a result of familiarization.Third, the results of this study were compared with the English data of Karmiloff-Smith(1981) in an attempt to resolve differences between her study and Bamberg’s regardingthe age at which children first evidence referential use of pronouns, and the exact nature102of the thematic advancement strategy which marks the first stage in the development ofpronominal reference.In addition to the three main goals above, the study also checked for correlation in thedata for various measures of development across subjects. This was done to determinewhether any one of these measures could serve as a specific indicator of developmentfor any other. Correlation between chronological age, mean length of clause and story“goodness” was measured. Further to these correlational measures, it was determinedwhether the data “split” with respect to performance on a key element of the thematicadvancement strategy for the youngest four and oldest four subjects.From his data and results, Bamberg proposed a theory of the developmental stagesthrough which children move in the acquisition of adultlike pronominal reference. Thefirst of these proposed stages was what he called a “global thematic advancementstrategy,” in which children preferentially choose a pronoun to both switch reference tothe protagonist of a story and to maintain reference to the protagonist. Simultaneously,children choose full nominal expressions to switch reference to other characters and apronoun to maintain reference to those characters. He found this strategy to be firstevidenced as early as 3;6.The design of this study replicated that of Bamberg (1987), with only a few minorchanges in design, to allow direct comparison of results. Bamberg did not collect oranalyze stories which resulted from th first (unfamiliar) condition, but stories from both103conditions were compared with his data in this study, recognizing that only the datafrom the second (familiar) condition was directly comparable. By replicating Bamberg’sstudy, the data in this experiment were collected in the same elicitation context, andtherefore any differences which might have been found in the nature or use ofpronominal reference could not be attributed to variations in design and method.Summary of ResultsComparison with Bamberg’s ResultsThe results showed general similarity between the stories of 3-year-olds Bambergcollected and the stories collected here. The children referred to the boy (the storyprotagonist) more often than the dog, and preferred the use of pronouns, regardless offunction, to refer to the boy. A strong preference to use nominals to refer to the dogwas not found. This was different from Bamberg’s data, and reflected the greatertendency of children in this study to maintain reference to the dog.Closer examination of the grouped data indicated evidence of a developing thematicadvancement strategy as described by Bamberg. The children chose significantly morepronouns than nouns to both switch and maintain reference to the boy. For referenceto the dog, there was a significant preference for pronouns to maintain reference, anda preference for nominals to switch reference at T1 (unfamiliar condition). These resultswere consistent with the thematic advancement strategy. However, at T2 the preferencefor nominals to switch reference to the dog was not significant, contrary to the thematic104advancement strategy. Despite this, there was still a significant difference at T2 (as wasfound at T1) between the number of pronouns used to switch reference to the boy andthe number of pronouns used to switch reference to the dog. There was no significantdifference found at either time between the preference to use pronouns to maintainreference to the two characters. Therefore, while the thematic advancement strategy wasnot as “strongly” in place for the children in this study (as evidenced by lack ofsignificant preference for nominals to switch reference to the dog at T2), the storiesdemonstrated trends toward development of the thematic advancement strategy. Theboy and the dog were definitely being treated differently with respect to pronominalreference, in a way that recognized the status of “protagonist” for the boy by creating forhim a unique system of reference. Therefore the children appeared to be showingsensitivity to aspects of textual coherence in a way similar to the youngest children inBamberg’s study. This is a confirmation of Hypothesis 1.Comparison Between T1 and T,No evidence was found in this study for a significantly greater ability of children toapply the thematic advancement strategy when they were familiar with the story as aresult of exposure to an adult model, and therefore Hypothesis 2 was not confirmed.This was true even though the nonreferential measures of narrative competence(percentage of clauses contributing to thematic advancement, mean length of clause, and“story goodness”) showed a trend toward the children telling shorter stories with agreater proportion of “contributing” clauses and inclusion of more story elements in thefamiliar condition. This lack of significant change in use of pronominal reference was105most likely due to a gap in competence and performance: the children may not havegained enough in competence from hearing the adult model to result in a change inperformance for this particular reference measure. As mentioned in the discussion, alonger and/or more intense period of exposure to the adult model may have resultedin improved application of the thematic advancement strategy.Joint ReferenceWhen analysing the data for evidence of the thematic advancement strategy, a lesserdistinction in the treatment of the boy and the dog was observed when joint referenceto the boy and the dog was included in the analysis. In particular, for switch referenceto the dog, a much stronger preference for pronouns to fill this function was found withinclusion of joint reference. This indicated that when making joint reference to the boyand dog, the children were (as a group) following the strategy for making reference tothe boy alone. It was concluded that an operational rule within the thematic subjectstrategy was:when the protagonist is part of a joint referent, treat the Joint referent as onewould treat the protagonist.Presence of such a rule was also supported when closer analysis of individual storieswas undertaken. It was found that half of the pronouns used to switch reference to thedog (an apparent violation of the thematic advancement strategy) were joint referenceswith the boy.106Comparison with Karmiloff-Smith (1981)The results of this study did not show evidence of solely deictic use of pronouns asKarmiloff-Smith (1981) predicted. Using the presence of point as a measure of deixis,the results showed that pronouns were not accompanied by point significantly morethan full nominals when Subject 3 was excluded from the data (an interpretation whichwas justified in the discussion). In contrast, other constructions that were truly deictic(e.g. There’s a X) were accompanied by point significantly more often than they were notaccompanied by point. Therefore, it was concluded that because pronouns and nounswere not being treated differently with respect to the use of accompanying point, andbecause within the narrative pointing was still a defining feature of clearly deictic terms,then pronouns could not be labelled a purely deictic term. This was taken as indirectsupport for the referential function of many pronouns in these narratives. Therefore, thefirst part of Hypothesis was not confirmed.It is recognized that other measures of deixis could have been applied (e.g. eye gaze,head movements), but for this analysis the results were in conflict with Karmiloff-Smith’sclaim that pronouns serve only deictic functions before age six. The results also plainlydiffer from those of Tomasello et al. (1984/85), who found a greater occurrence of pointwith pronouns than with nouns for three-year-olds in a nonnarrative discourse task.This implies a change in function of pronouns for children of this age group whenplaced in the narrative context, which would support a linguistic sensitivity to the genreof “narrative” on the part of the children.107The results of this study did not show an exclusive preference to reserve pronouns inclause-initial position for reference to the protagonist as would be predicted fromKarmiloff-Smith’s description of her thematic subject strategy, and therefore did notconfirm the second part of Hypothesis 3. Only 51% of the pronouns in initial positionmade reference to the boy. Of those pronouns in clause-initial position which served toswitch reference, 65% of these were to the boy. While this does not support exclusivereservation of the clause-initial pronoun for reference to the boy, it still shows atendency to refer by pronoun to the boy in clause-initial position more than for anyother character.1Furthermore, when clause-initial position for pronouns was looked at from theperspective of reference to the boy, it was found that 86% of all pronominal referencesto the boy were in clause-initial position. In addition, 56% of the pronouns used toswitch reference to the boy were in initial position. This implied that the clause-initialposition was in fact special for the protagonist of the story, even if not to the extentpredicted by Karmiloff-Smith. This was an expected result, because children will createfocus around the protagonist of a story, and therefore often choose him to be the agentof clauses, especially after he has been reintroduced in an episode (Bennett-Kastor, 1981).The relatively low preference to reserve clause-initial position for the boy (with respectto Karmiloff-Smith), and the lower preference of that position to switch reference to theAs mentioned in the discussion, Karmiloff-Smith did not give actualnumbers for the percent of clause-initial pronouns which referred to theboy, but the implication was that the preference was greater in her studythan that found here.108boy, was attributed to the number of episodes in which the boy was not the main focus(and therefore another character was chosen as agent of the action) and in which the boywas reintroduced as the recipient of an action (and therefore the grammatical object).The clause-initial data also revealed that the boy was being treated differently than othercharacters with respect to the pattern of introducing (or reintroducing) and maintainingreference. Of all cases of clause-initial maintain reference, 36% were “local,” meaningthat a character was introduced by a full nominal and then maintained by a pronounwith no intervening referents. The fact that there were no cases of maintain reference(local) for the boy indirectly implied that none of the cases in which a full nominal wasused to introduce or reintroduce the boy was part of such a localized pattern ofintroduction and maintenance of a character. This again supports the notion that, inlinguistic terms, the protagonist (a feature of the narrative context) is treated differentlyfrom other characters in the narrative.Karmiloff-Smith concluded from her study that children of three years do not have theability to create the intersentential linguistic links necessary to create narrative coherence.As the discussion above implies, the present study (like Bamberg’s) did find evidenceof the developing ability to create local cohesion and textual coherence. Further to theconclusion that the children used pronouns in a nondeictic manner, the data were foundto “fit” with Bamberg’s same-age German data, and supported the use of a thematicadvancement strategy by 3-year-olds. In addition, errors and self-corrections made bythe children further revealed their sensitivity to the “special” quality of the story109protagonist. These errors and self-corrections also revealed other forms which filledcohesive functions and the ability of at least one child to generalize the entire searchmotif through the repetition of one phrase (“they looked everywhere”). Therefore, it wasconcluded that, although story schemas were not complete and deictic constructionswere common in the stories of these children, a sensitivity to aspects of the narrativewhole and the initial capability for creating textual coherence did exist.Individual DataFurther investigation of individual stories indicated that many of the apparent“violations” of the thematic advancement strategy were either attributable to individualnarrative style of some children, to the use of joint reference (as discussed above) or toinconsistencies created by the coding system. This “second look” at the data indicatedthat one subject (Subject 3) was generally using pronouns deictically and was still in theearly stages of developing referential pronoun use. Two other subjects (Subjects 4 and7) seemed to be at a more advanced point in this development, as they frequently usedpronouns to serve referential functions, but still exhibited the use of deictic expressionsto serve these functions elsewhere in their narratives. Other aspects of individualnarrative style which affected the use of the thematic advancement strategy includedgiving names to characters (i.e. proper nouns) and using story conventions. These onlyoccurred early in the second tellings and were most likely the direct result of the adultmodels children received. They were presented early in the narratives and were quicklyabandoned.110The analysis of individual stories also led to the observation that some children appliedthe thematic advancement rule:use pronouns to switch reference to the protagonistwhen switching reference to the dog in episodes where the dog was the main character(and therefore the “doer” of the actions). It was concluded that these children had notyet completely developed the notion of “protagonist” as enveloping the whole narrative.However, there were indications that this application of reference rules normallyreserved for the boy was “uncomfortable” for the children. A substrategy was found inthe stories of some children whereby the inanimate pronoun “it” was used to switchreference to the dog in these episodes despite the heavily favoured use of “he” to referto the dog elsewhere in the story. The number of these examples was few, but it wasconcluded that the presence of such a substrategy to maintain the distinction betweenthe dog and the boy was further indication of developing sensitivity to “global” aspectsof the narrative.Across Subject MeasuresThe analysis which investigated correlation between the measures of chronological age,mean length of clause and “goodness” of story showed no significant correlation betweenany of these measures. Although each is separately considered a measure of linguisticand/or narrative development, these results indicate that, in this narrative context, nonemay serve as a developmental cue for any other. The second analysis across subjectswith respect to chronological age investigated the relationship between chronological ageand the preference to choose pronouns to switch reference to the boy (i.e. to apply the111thematic subject strategy). The oldest four subjects did use pronouns to switch referenceto the boy significantly more of the time than the youngest four subjects. This “split” inthe data confirms the observation that the thematic subject strategy is still developingin 3-year-old children.Overall Summary of ResultsTo summarize the results, this study provided evidence that English speaking childrenbetween the ages of 3;2 and 3;9 demonstrate the developing ability to create textualcoherence through pronominal reference in a manner which supports Bamberg’sthematic advancement strategy. It did not provide evidence for purely deictic use ofpronouns or the inability of children to create any level of textual coherence found byKarmiloff-Smith (1981) for English-speaking four-year-olds. This study further disputedKarmiloff-Smith’s results when exclusive preference for the clause-initial pronouns to bereserved for reference to the protagonist was not found. It is proposed here that thediffering results (both in the age at which children first use pronouns as referencedevices constrained by the creation of textual coherence, and the range of grammaticalroles assigned to the protagonist) can be aftributed to variation in the experimentaldesign.Explanation of Differences Between StudiesAlthough Karmiloff-Smith also elicited the narratives in her study from a picture book,the book had far fewer picture frames than Frog, Where Are You?, and also had fewercharacters and episodes. According to researchers such as Liles (1988, 1993) and112Peterson & McCabe (1983), this will result in shorter and less complex narratives.Indeed, examples of elicited narratives Karmiloff-Smith did provide in her paper weretypically of six clauses in length, as opposed to an average length of over fifty clausesfor both tellings of Frog where are you? in this study. Because “frequency of use [of theunit being investigated] is the typical measure to distinguish one population fromanother” (Liles, 1993: 877), it may be that developmental patterns in the narratives ofchildren under six were lost due to infrequent (or absent) occurrence. Therefore, the useof pronouns as referential devices which formed patterns in the narratives of this andBamberg’s study may have been perceived as aberrations in the data by Karmiloff-Smith.They may have been there, but not in sufficient numbers to be “noticed.”Further, the protagonist in the story used by Karmiloff-Smith is involved throughout thestory and is the focus of all events (even when not the agent). Unlike Frog, WhereAre You?, there are no episodes within the story in which the protagonist does not takepart, and therefore no opportunity for the protagonist to be reintroduced following suchan episode. Because there is little opportunity for the “focus” of the narrative to shiftbetween characters (as described by Bennett-Kastor, 1983), it is natural that pronounswhich referred to the protagonist were the agents of actions, and therefore grammaticalsubjects in utterance-initial position. Therefore, it is proposed that Karmiloff-Smith’sstories from children under six years of age were limited in data pointing to referentialpronoun use and developing strategies for creating textual coherence. They were alsolimited in the range of characters referred to by utterance-initial pronouns. Theselimitations were most likely due to the lack of length and complexity in the book used113for elicitation of the narratives. Karmiloff’s claims are therefore not warranted, given thelimitations due to experimental task and materials.The importance of method and design in comparing the results of studies which examinethe same functional unit (like pronominal anaphora) is further reflected in Orsolini’sresults (1990). In her study, she found support for aspects of the thematic advancementstrategy proposed by both Bamberg and Karmiloff-Smith (i.e. preferential use ofpronouns for switching reference to the protagonist of a story). She found evidence ofthis strategy in the narratives of four-year-old children, but only in 14% of thesenarratives. She explained this low incidence of the strategy as the result of a stilldeveloping episodic structure, an explanation well-supported by the literature. Thisresult is somewhere in between the results of my study (and Bamberg’s) and those ofKarmiloff-Smith, in terms of “strengtW of the thematic advancement strategy. While thethematic advancement strategy was observed, it was not as well-established at four yearsof age as Bamberg found. However, Orsolini’s method used a televised, verbal narrativefrom which children retold the story later with no visual support. Therefore, the contextin which the children gave their stories provided less support for the content andorganization of story episodes than in the case of this (and Bamberg’s) study. The lengthof the model narrative, however, was more comparable to Bamberg’s story than toKarmiloff-Smith’s, The result of Orsolini’s analysis, then, seems to be a compromisebetween greater opportunity for the form/function pairings of the thematic strategy tobe evident (due to story length), and more effort on the part of the child being directedat remembering and organizing the story (due to elicitation context).114The Scope of the Present StudyThis study did not confirm or dispute the developmental sequence in the acquisition ofpronominal reference proposed by either Bamberg or Karmiloff-Smith. Rather, thisstudy confirmed Bamberg’s finding that children as young as three years of age can usepronouns referentially in functions which reflect attention to the advancement of themewithin a narrative, and in doing so aid in the creation of textual coherence. This studyalso showed that the first stage of referential pronoun use by English-speaking childrencomes at approximately the same chronological age and in the same form - the thematicadvancement strategy - as by German-speaking children (Hypothesis 1). This is also aconfirmation of Karmiloff-Smith’s proposed first stage (despite differing in the age offirst appearance) although exclusive reservation of clause-initial pronouns for referenceto the protagonist was not found.Future ResearchIn order to confirm Bamberg’s model of the developmental stages and the timing of theirappearance in acquisition of adultlike anaphoric pronominal reference, the mostconvincing method would be a longitudinal study of these same children. By askingthem to tell a story of comparable length and complexity to Frog, Where Are You? atintervals over their development, using the same methods of elicitation and analysisused in this study, a valuable pool of data would be collected. By doing so, even thesubject variability found in cross-sectional studies would be removed. The difficultiesof maintaining such a study over time are recognized; however, due to the sensitivity115of narrative production to elicitation context, a longitudinal study of this type would beideal.Further, this study did not take into account the influence of adult models on thenarratives produced by children. The presence of proper nouns and story conventionsas a result of exposure to the adult model were mentioned, but not formally assessed.Further study into the effects of the “quality” of adult stories (in terms of completeness,length, complexity of language used) on the story retellings of children would bevaluable information. It would be possible to do such an analysis for this data set, astapes of the adult models were collected for most subjects, but such analysis was beyondthe scope of this study.As mentioned before, it would also be interesting to determine under what conditions(if any) manipulation of the length and/or intensity of exposure of children to a varietyof controlled adult models would result in a significant improvement in their applicationof the thematic subject strategy. This would provide valuable information not onlyabout the “cognitive effort” children are distributing between aspects of narrativeproduction, but also more about the contexts in which story retellings are elicited.Finally, because children as young as 3;2 were able to apply the thematic advancementstrategy to their (incomplete) narratives in, it would be of interest to repeat this studywith a group of children of a younger age range. More narrative data would becollected in which children were not using pronouns referentially, and were truly unable116to manage the continuity of characters and events throughout the narrative eitherconceptually or linguistically. More data which showed that children were “caught inthe middle” of transition to the first stage of pronominal reference would also becollected. Close analysis of this data would further elucidate processes and strategieschildren are employing in the earliest stages of referential pronoun use.Concluding RemarksThis study gives credit to these young children for their developing abilities to managethe narrative production task from a variety of perspectives. In particular, it providedevidence that 3-year-old children do pay attention to aspects of local cohesion andtextual coherence in their narratives, as was shown here for the use of pronouns asreference devices. What is more, these children were able to demonstrate theirknowledge of narratives and linguistic structure and function while telling a story, a taskthat is far from simple.This study demonstrates the importance of form/function approaches in examining thedevelopment of the referential system, and indeed language development in general.By investigating the functions served by referential forms in the narratives of this study,the presence of a relatively sophisticated strategy for creating both local cohesion andtextual coherence was demonstrated. This strategy was present despite the apparentinability of these children to tell complete, complex stories. The potential ability of sucha form/function approach to demonstrate developing linguistic and discoursecapabilities which might otherwise be overlooked must be kept in mind when examining117the narratives of young children, in addition to the importance of the experimentalcontext from which the narrative data are drawn.118BIBLIOGRAPHYBamberg, M., (1987). The Acquisition of Narratives. Berlin: Mouton de Gruyter.Bennett-Kastor, T. (1983). Noun phrase and coherence in child narratives. Journal of ChildLanguage, 10, 135-149.Berman, R., (1988). On the ability to relate events in narrative. Discourse Processes, 11,469-497.Berman, R. & Slobin, D., Bamberg, M., Dromi, E., Marchman, V., Neeman, Y., Renner,T. & Sebastion, E. (1986). Coding Manual: Temporality in Discourse. Revised edition.Cognitive Science Program. Univ. of Calif., Berkeley.Brighouse, J. (1990). Coast Salish Children’s Narratives: Structural Analysis from ThreePerspectives, Unpublished Master’s Thesis, University of British Columbia.Brown, G. & Yule, G. (1983). Discourse Analysis. Cambridge: Cambridge UniversityPress.Emslie, H. & Stevenson, R. (1979). Developmental aspects of communication: Youngchildren’s use of referring expressions. In P. Worth (ed), Conversation and Discourse:Structure and Interpretation, London: Croom Helm.Chafe, W., (1990). Some things narratives tell us about the mind. In Britton & Pellegrini,A. (eds), Narrative Thought and Narrative Language. Hillsdale, N.J.: Lawrence ErlbaumAssoc.DuBois, J., (1980). Beyond definiteness: The trace of identity in discourse. In W. Chafe(ed), The Pear Stories. New Jersey: Ablex Publishing Corp.Halliday, M. & Hasan, R., (1976). Cohesion in English. London: Longman.Hickman, M., (1980) Creating referents in discourse: a developmental analysis oflinguistic cohesion. In J. Kreiman and A. Ojeda (eds), Papers from the Parasession onPronoun and Anaphora. Chicago: Chicago Linguistic Society.Hudson, J. & Shapiro, L., (1991). From knowing to telling: The development ofchildren’s scripts, stories and personal narratives. In A. McCabe & C. Peterson (eds),Developing Narrative Structure. Hiisdale, N.J.: Lawrence Erlbaun Assoc.Karmiloff-Smith, A., (1979). A functional approach to child language. Cambridge Studiesin Linguistics, 24. Cambridge: Cambridge University Press.119Karmiloff-Smith, A., (1981). The grammatical marking of thematic structure in thedevelopment of language production. In W. Deutsch (ed), The Child’s Construction ofLanguage. London: London Academic Press.Liles, B., (1993). Narrative discourse in children with language disorders and childrenwith normal language: A critical review of the literature. Journal of Speech and HearingResearch, 36, 868-882.Martin, J., (1985). Process and text: two aspects of human semiosis. In J. Benson & W.Greaves (eds), Systemic Perspectives on Discourse, vol.1. New Jersey: Ablex PublishingCorp.Mayer, M., (1969). Frog, Where Are You?. Hong Kong: South China Printing Co.Merritt, D. & Liles, B., (1987). Story grammar ability in children with and withoutlanguage disorder: story generation, story retelling & story comprehension. Journal ofSpeech and Hearing Research, 30, 539-552.Pellegrini, A. and Galda, L, (1990). The joint construction of stories by preschoolchildren and an experimenter. In Britton & A. Pellegrini (eds), Narrative Thought andNarrative Language. Hillsdale, N.J.: Lawrence Eribaum Assoc.Peterson, C. and McCabe, A., (1991). Linking children’s connective use and narrativemacrostructure. In A. McCabe & C. Peterson (eds), Developing Narrative Structure.Hillsdale, N.J.: Lawrence Erlbaum Assoc.Schiffrin, D., (1990). Between text and context: Deixis, anaphora and the meaning ofthen. Text, 10(3), 245-270.Snow, C. and Goldfield, B., (1981). Building stories: The emergence of informationstructures from conversation. In D. Tannen (ed), Analyzing Discourse:Text and Talk.Washington, DC: Georgetown University PressStein, N., (1988). The development of children’s storytelling skill. In M. Franklin & S.Barten (eds), Child Language: A Reader. Oxford: Osford Univ. Press.Stein, N. & Glenn, C., (1982). Children’s concept of time: The development of a storyschema. In W.J. Friedman (ed), The Developmental Psychology of Time. New York:Academic Press.Tomasello, M., Anselmi, D. & Farrar, M.J., (1984/85). Young children’s coordination ofgestural and linguistic reference. First Language, 5, 199-210.Wales, R. (1996). Deixis. In P. Fletcher & M. Garman (eds), Language Acquisition. Secondedition. Cambridge: Cambridge University Press.120Westby, C., (1984). Development of narrative language abilities. In C. Wallach & K.Butler (eds), Language Learning Disabilities in School-Age Children. Baltimore, MD:Williams and Wilkins.Wiese, B., (1983). Anaphora by pronouns. Linguistics, 21, 373-417.121APPENDIX AFROG, WHERE ARE YOU? (MAYER, 1969): PICTURE BY PICTURE DESCRIPTION1. Boy, dog and frog are in bedroom;boy and dog are watching frog who is in a jar.2. Boy and dog are asleep in bed; frog is stepping out of the jar.3. Boy and dog are awake and look at the empty jar from the end of the bed.4. Boy looks in one of his boots;dog sticks his head in the jar.5. Boy and dog are at window; boy is calling and dog has his head stuck in the jar.6. Dog is falling from window ledge; boy is watching him fall.7. Boy is down on ground below window holding dog; dog is licking boy’s face andthere is broken glass on the ground.8. Boy is calling towards forest; dog is sniffing at a line of bees coming from a hiveat the edge of the forest.9. Boy is calling into hole in ground; dog is barking at bee hive.10. Boy is holding his nose as if in pain; a little animal is at hole entrance; dog isleaning against bee hive tree.11. Bee hive is on the ground and the bees are exiting en masse; boy is up a treelooking in a hole.12. Boy is on his back on the ground; an owl is at the entrance to the hole in the tree;the bees are chasing the dog, who has run past boy.13. Boy is holding his hand above his head as if to fend off owl, who is flying abovehim; boy is at the bottom of a large rock.14. Boy is calling from the top of the rock; he is leaning on some things behind therock that look like branches.12215. Boy is on top of a deer’s head between its antlers (the “things” of the previouspicture turn out to be these antlers); dog is almost entirely behind the rock wherethe deer is.16. Deer is running towards a cliff with boy on his head; dog is running beside deerwatching boy.17. Deer stops at edge of cliff; boy and dog are falling over cliff towards a body ofwater.18. Boy and dog splash into water.19. Boy is sitting in the water with dog on his head; boy has hand to his ear as iflistening to something.20. Boy is leaning against a log and saying “Sh” to dog who is now in the waterbeside boy.21. Boy and dog look over to the other side of the log.22. Boy and dog are on top of the log; two adult frogs are on the other side.23. Nine baby frogs have joined the adult frogs (who are looking like proud parents).24. The frogs are on top of the log facing towards the water; boy and dog arewalking through the water away from the log; boy, who has one of the babyfrogs, is looking back towards the family of frogs and waving.(Copied with permission from Brighouse, 1990.)123APPENDIX BSample TranscriptN: ResearcherS: Subject 4First Telling, Subject 4N What’s happening?S (Urn urn) (the) The dog looked at the frog.N Mrnhmm.S (And then the) And then this boy sit dowxi.S and watched.S He didn’t know that.N (laughs} OK.Then what happened?S Mmm, (Then he) (And then he) And then he cried.S {taking off boots) I’m taking my boots off.N Are ya?OK.124All right.What’s happening?S (Urn) (urn) The frog is getting outS and going.S and then he’s putting the (Urn) the boy’s (sh>) slippers on.N Ohishe?And then what?S (urn) Then the boy cried.N Mrnm.S And he didn’t know that happened.N Oh.Do you wanna help turn the pages?S Yah.S And then he says, “1-1mm? Where’s that frog?”.N {laughs}N Yah!S {pointing to painting in room) Who’s is that picture?N I don’t know, it looks like it says Ellen.So then what happened?What’s going on?S He didn’t know that.N Hmmm?S He didn’t know that.125N What?S He didn’t know that.N Oh, you help turn the pages.S Nope.N Yah.S I don’t want to.N OK you tell me when to turn the page then OK.S No.N OK what’s happening?S And now he went out.N Mmhmm.S And then what happened is someone came out of (the) the bush.N Really?S Let’s see.N OK.S Now.(is)N What’s going on?S (Urn) The wind is blowing.S and then the trees are breaking.S and getting like a monster.N And what?S (Urn) the trees are blowing.126S and they are getting like a monster.N Oh.S And it’s night, right?N You think so?(iS)What’s going on?S (urn) It’s night!N Mrnhrnm.S And the dog is angry.S (And h>) He’s calling (is) for that boy.N Isee.S Then what happened he said, “Huh?”S (Is the) (Is that boy) (Now the) Now that boy says, “Is there a animal here?”(S goes under the table;negotiate S finishing the story: 21 s)S Now turn the page.N OK what’s happening?S (changing position) I’m going to (st) (st) (n) stand up.N OK how about you kneel.S OK.N OK there we go.All right.S (mm) And then he climbed up the treesS and went in a hole.127N Really!S And now turn it.S Let’s look. {whispered)S There’s a owl!N Oh!And what else?S The end (?)S And there’s baby owls he has.N Mmhrnrn.S And the Mommy owl.N Really, and what else is going on?S (Urn) The dog is running away.N Mmhmm.S (And th>) And then (s>) (urn urn) crocodiles get hirn.S Let’s see what happened.S Mrnrn!N Oops.N Mrnhmm, what happened?S (urn urn urn) (urn) The snow (gr>) growed up (to the) (he) (to the) to the cloud.N Oh wow.Then what?S Then he would get scaredS and run back home.128N Now what’s happening?S (Urn) The dog growed up like a deer.(is)S (Uh) I mean there was a mean bad guy deer.N Whoa; what happened?S And the dog ran away.S (And that) And the boy caught on the deer.N Isee.Then what?S He fell off.S and the dog (is) on a cliff. [kllft]N Oh my goodness.S And it’s a cliff.N Mmhmrn.S And they are gonna fall in the water.S Yah I’m right!N (S is restless) Here come on over here.(laughs)Then what?S Aah! He thought it called.S Turn the page.S Look what happened. (whispered)N What’s happening here?i29S (Urn) He went in a wood.S And he cried and cried into a wood.S And (didn’t) (d>) didn’t know where he was.N And what?S And the dog and the boy went into this wood.N Oh, the wood?S Yah.S In to hide.N Isee.S That’s where they wanted to go.S (Now the) And then they said, “Shoof!”.N They said>S “Shoof!”N Oh.What’s happening here?(is)S Frogs came.N Isee.S They were happy.5 (And then this) There (w>) was other babies.S What is this? (Pointing to FM equipment)N Let’s finish the story; it’s almost done.(is)i30S And then (urn) the frogs jump.S Still stayed in the home.S And then they went back home.N Isee.S And turn the page.S That’s all!131Second Telling, Subject 4S It’s night now.N You’re right.S And then this guy watched him.N Mmhmm.S And then one morning he said, “The Froggy is gone!”N Yah.S (He) The doggy looked in the frog jar.N Mmhmm.S And this guy looked in the boot.S And no froggy.N No? Oh no!OK.S Turn the page.N Turn the page.S And now that guy look this guy.N Mmhmm.S And then this guy broke the glass.N Mmm.N Right.S And then he said, “Froggy where are you?”N Right, and what else?S And then he looked in this other hole.132S And then he looked in this hole.S And then there was an owl.S And a beehive.S Or a squirrel hive.N Oh.S Look what I (f>) saw!S And he bit his nose.N Oh! Ouch!S The beehive.S And then (this) the tree fell down.S And then all the bees chasing the dog.N Oh, look.S They said, “Froggy, where are you?” again.N Uhhuh.S And it fell down.S And then the thing fell.(is)S and it went Boom!N tlaughs} Boom.S And (the) this guy said, “Froggy, where are you?”N Yah?S And then a owl cameS and snapped he.133N And what?S And some owls snap people.N Ohlsee.S And then this guy ran away so fast.N Uh huh.S (And th>) And he holded on to the branches.N Uhhuh.S and they weren’tS And it wasn’t a branch!N Oh! Oh goodness!S Ohno.S And then this guy throwed him down on the cliff.N Oh! Oh my goodness.S And then (this urn) this guy laughed.N Hmm! Ha ha ha!S And this guy said, “Whoaaa!”N (laughs)S This guy said, “Woof woof!”N (laughs}S That’s what they said.N Oh!S And then this guy said, “Shhh!”S And then (the froggy) the froggy said, “Ribbit, ribbit”134S XXXXN Mmhmm.S He said, “Shhh!”S And he found him!N Ohh!(2s)Then what?S One, two, three, four, five, six, seven, eight, nine, ten, eleven.N And what’s happening?S One (is) jumped on this guy’s hand.N Mmhmm.S And now they went away.N Ohlsee.135APPENDIX CSubcategorized Form/Function Coding PairsTable Cl. Numbers and percentages of reference to the boy and to the dog accordingto specified form and function, T1 and T2 (joint reference to boy and dog included)SWITCH MAINTAIN SUMT1 (%) T2 (%) T1 (%) T2 (%) T1 (%) T2 (%)To the boy:Nominal 10 (13) 2 (6)-- (0) 1* (1) 14 (13) 11 (7)3$1# 6$Pronominal 43 (50) 65 (57) 26 (31) 36 (28) 91 (81) 131 (85)3a 2a 9# 7#10# 21#Ellipsis 3 (2) 4 (2) 4 (4) 8 (6) 7 (6) 13 (8)1#Sum 73 (65) 102 (65) 39 (35) 53 (34) 112 155To the dog:Nominal 35 (42) 20 (21)-- (0) 1 (3) 40 (42) 27 (24)4$ 4$ 1$1# 1*Pronominal 12 (27) 18 (36) 17 (28) 33 (37) 51 (55) 81 (73)3a la 9# 8#l0# 21#Ellipsis-- (0) -- (0) 3 (3) 2 (3) 3 (3) 3 (3)1#Sum 65 (69) 64 (58) 29 (20) 47 (43) 94 111joint reterence to boy and dog (e.g. t Ley, them etc.)* nominal reference to last mentioned character at beginning of new event orpagea appropriate use of pronominal reference in switch reference positionp proper noun$ deictic term (e.g. this/that X, a(n) X <after character previously established>)#136Table C2. Numbers and percentages of reference to the boy and to the dog accordingto specified form and function, T1 and T2 (joint reference to boy and dog excluded)SWITCH MAINTAIN SUMT1 (%) T2 (%) T1 (%) T2 (%) T1 (%) T2 (%)To the boy:Nominal 10 2-- (0) 1*(1) 13 (14) 11 (9)3$ (14) 2p (8)6$Pronominal 43 65 26 (28) 36 (29) 72 (78) 103 (82)3a (50) 2a (53)Ellipsis 3 (3) 4 (3) 4 (5) 8 (6) 7 (8) 12 (9)Sum 62 (67) 81 (64) 30 (33) 45 (36) 92 126To the dog:Nominal 35 20-- (0) 1 39 (53) 27 (33)4$ (53) 4$ (30) 1$(3)1*Pronominal 12 18 17 (23) 33 (41) 32 (43) 52 (64)3a (20) la (23)Ellipsis-- (0) -- (0) 3 (4) 2 (3) 3 (4) 2 (3)Sum 54 (73) 43 (53) 20 (27) 38 (47) 74 81I. nominal reterence to last mentioned character at beginning or new event orpagea appropriate use of pronominal reference in switch reference positionp proper noun$ deictic term (e.g. this/that X, a(n) X <after character previously established>)137


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items