Identification in narrative: Evidence for conversational processes

Nelson/Denman , 9:00 - 10:00 A.M.

An impressive number of studies have now demonstrated that writing about emotional upheavals can produce improvements in physical and mental health. Using a text analytic strategy, we have been exploring the features of word use that predict these health changes. Further, language use in the writing paradigm predicts changes in the ways people naturally talk to others in the weeks after the study. Recent findings about using language and word choice as a fingerprint of personality and social situations will be discussed.

Denman Room, 10:30 A.M. - 12:10 P.M.

This study assessed the viability of developing a reading skills test based on thinking aloud. Good and poor readers thought aloud while comprehending two Chinese folktales. Both traditional, qualitative analyses and nontraditional quantitative analyses (i.e., Latent Semantic Analysis) of the content of the think-aloud protocols were conducted in order to expose differences between good and poor readers in terms of the content of their think-aloud protocols. The results indicated that latent semantic analysis may provide an important basis for assessing reading skill based on the content of think-aloud protocols.

We examined the extent that above and below-average readers allocated resources to building the situation model of descriptive text across two readings. One group was instructed to read in order to take a test; a second learned a diagram of the text referents before reading; a third was told to read for understanding. The test and diagram facilitated the on-line construction of the situation model for the better readers, but inhibited the construction for the poorer readers.

In this study, we investigated whether the updating process is sensitive to readers' prior knowledge and task demands. We distinguished participants according to their prior knowledge and we assessed whether they integrated new incoming information via two tasks that probed the same information but differed on their temporal occurrence during the experiment. Results confirmed that the updating process is composed of two temporal components which are differently affected by the specificity of prior knowledge and by task demand.

We argue that identifying with a character in literary narrative occurs because the narrator is treated as a conversational participant. This entails that readers draw narrational implicatures on the assumption that the narrator is rational and cooperative, and as a consequence, attribute aspects of their own knowledge and experience to the narrator. Consistent with this analysis, less identification occurs when the narrator’s thoughts and attitudes are explicitly justified, making implicatures unnecessary.

Nelson Room, 10:30 A.M. - 12:10 P.M.

In our analysis of a corpus of spoken dialogues we show that in spontaneous speech only 45% of pronouns and demonstratives have NP antecedents. For the remaining anaphors we provide a classification into three types, which are reliably marked by human annotators -- discourse-deictic reference to events and propositions, vague reference to the general topic of discourse, and inferrable-evoked reference. We furthermore show that these types are mainly restricted to spoken language and discuss implications of this.

We report first findings in a study of the use of lexical subjects in conversation. We find that speakers who choose to conflate the discourse functions of referent-establishment and predication through the use of a lexical subject nevertheless constrain hearer processing load by ensuring that these lexical subjects denote referents which are ACCESSIBLE or ANCHORED to the discourse context. This claim is based upon morphosyntactic coding trends in the Switchboard corpus.

This study maps the naturalistic emergence and evolution of topic production (or topicalization) in an experimental relationship of five conversations. The analysis identifies the correspondence of topical elements between topics and subtopics, both adjacent (topic drift, Hobbs, 1990) and non-adjacent (topic coherence). The resulting statistical and graphical profiles of topical flow reflect characteristic and quantifiable differences among conversations. Applications of this method mapping topical flow are discussed.

Recent evidence suggests that parent-child talk influences the development of children's social understanding. To evaluate this general finding, variables from videotaped parent-child interaction were derived and related to independent measures of three - to - 5 - year-old children's understanding of emotion. Aspects of talk were positively related to children's overall measure of emotion understanding. Partial evidence supports the hypothesis that social interaction is a necessary component in the development of children's social understanding.

Denman Room, 12:30 - 1:00 P.M.

Belief change may be compared to the "deep processing" theory in cognitive psychology described by Craik and Lockhart (1972). While research on beliefs has grown in the US, this work has not been related to emerging democracies and the cultural and conceptual shifts in beliefs found in such worlds. Since 1991, with the end of Communism, political reforms have taken place to move Russia and The Ukraine into democratic republics. Russians and Ukrainians are routinely engaged in discourse processes which involve questionning and reflecting about cultural and conceptual beliefs. This slide presentation and analysis of post-Soviet worlds is based on my 1998 visit to Russia and The Ukraine. Using interviews, text accounts by journalists, and diary notes, as well as recent theory of belief change, the presentation considers shifts in belief and their influence on identity formation and literacy.

Denman Room, 1:45 - 3:00 P.M.

This study examines the relationship between reading skill and working memory (WM) capacity. Previous correlational research has indicated that skilled readers have greater WM capacity. We examined the effects of strategy use on WM task performance to determine whether metacognitive strategies contribute to both WM capacity and reading skill. Our overarching goal is to determine whether strategy use accounts for correlations between reading skill and WM capacity. Results presented here support that hypothesis.

Think-aloud protocols were analyzed to investigate the cognitive processes that readers with high and low working memory (WM) capacities employ when reading for study versus entertainment purposes. All readers produced more paraphrases and backward inferences when reading to study and editorial comments when reading for entertainment. High WM readers produced more elaborative inferences and, when reading to study, metacognitive remarks than did low WM readers. Thus, readers adapt processing to reading purpose but high WM readers engage in processes that go beyond basic comprehension.

Nelson Room, 1:45 - 3:00 P.M.

Children develop an understanding of mind; they learn the circumstances for the correct use of mental state terms. To examine this process, mother-child dyads were videotaped as they made up a story. Analysis of the videotapes revealed that mothers initiated most talk of mental states; mental states were named, described, explained, pointed to, and mimicked orally and facially; and children typically acknowledged mother's mental state talk, but at times contested it or demanded clarification.

Denman/Nelson Rooms, 3:25 - 5:05 P.M.

 Functional neuroanatomy of narrative comprehension with and without words

Communicating degrees of uncertainty, levels of risk associated with hazards, and quantities in general is a ubiquitous activity, undertaken by laypersons and experts alike. This paper concerns the effect of expressing quantities in different ways, including through number and through natural language. Our claim is that whichever way quantities are expressed, it is impossible to do it in a neutral way, and that this fact results from the way in which the language of quantity is structured to reflect a process of argumentation.

Four experiments demonstrate consistently that the availability of instrument concepts in comprehenders' working memory is affected by (1) the temporal contour of the action involving the instrument and (2) the instrument's availability for use in the situation. Specifically, the availability of the instrument in the situation affects its accessibility to the comprehender, but only if the action is described as ongoing not when it is described as punctual. Analyses of reading times yield a similar pattern, except that the instrument's availability now also shows a reliable effect for punctual actions, but this effect is reliable smaller than that for ongoing actions. The implications of these results for theories of situation-model construction will be discussed.

Ballroom Foyer, 8:00 P.M. - 10:00 P.M.

We investigated whether readers detect inconsistencies relative to the spatial and emotional dimensions when provided with prior knowledge in relation with these dimensions. We assumed an effect of prior knowledge regarding the monitoring of these dimensions as well as the detection of inconsistencies. Our main results showed that although the spatial dimension is more difficult to monitor than the emotional dimension, it allows the reader to create a stronger situation model when faced to inconsistencies.

The extent in which illustrations influence the comprehension and representation of text information in memory was examined. Participants either read passages in which target sentences were or were not embellished with an illustration depicting what the text described. The participants' recognition for the targeted sentences were then tested either immediately following the presentation of the texts or one week later. The results most strongly supported a situational effect, whereby the illustrations enhanced participants' memory for the situation being described in the text more so than for other types of representations.

The effects of narrative characteristics on thematic inferences were investigated. In Experiment 1, a self-paced reading-time method demonstrated that appropriate title facilitated comprehension of the theme of a text. In Experiment 2, results indicated that generation of thematic inferences depended on the interaction between the central goal of the protagonist and the outcome of a text. These findings are consistent with the constructionist theory in which the theme is constructed during reading by integrating crucial information.

The purpose of this paper is to present background information on interconnections between reading and writing and to speculate about the nature of their potential relations. Three theoretical models are examined and a global model of comprehension/production of written texts intended to lay down the basis for posing a common knowledge between both competencies will be outlined. At the same time, some suggestions for the development of the field will be made.

The relationship between argument structures and meaning has been acknowledged in various literatures, but the source of this relationship is far from clear. One approach proposes that this relationship is the result of verbs with a particular meaning projecting a particular argument structure. A contrasting approach holds that argument structures exist as independent linguistic units with their own semantics. In a series of experiments, we present evidence that supports the latter approach to argument structure.

This paper presents a corpus-based discourse account of the constraints on the occurrence of null elements in Hindi. It will be shown that an adequate account of the phenomena cannot be provided in terms of a theory of pro-drop tied in with verb agreement. Furthermore, the study will also contest previous attempts to explain the phenomena from a discourse perspective, and provide a proper account of null elements in the language in terms of "Centering Theory", which provides a strict definition of the notion of the "topic" of the utterance.

Genre is socially constituted through shared knowledge making practices of novice and experts. Social pragmatics is an empirical method which investigates conversation under the rubric of talk-in-interaction. Ten students (novices) and their instructor (expert) were taped recorded. The tape was analyzed for structures that constitute genre, expert and novice. Though a preliminary study, expected structures were confirmed. Significantly, the social pragmatic method proved an appropriate and successful method for analysis.

This study explores psychological features of topic production (topicalization) in conversation, where topical elements may variously divide, combine, disappear, resurface, and transmute. This ephemerality and mutability is demonstrated in a top-down analysis of a conversation, pointing to the role of speakers‚ internal (unspoken) topicalization. Next, this phenomenon, and its impact on the progress of the speakers‚ relationship, is examined across a series of conversations, illustrating the confluence of internal and external topicalization.

This current study is intended to give an account of the types of papers written by a sample of seventh graders of Spanish state schools in Valparaiso, Chile. The most common textual typology are the copy of other writings and the answers given to questionnaires; the least common are the persuasive texts and the schematic representations of written discourse. The detection of the typology is complemented by the inquiry of the quality of the same typology.

The aim of this investigation is to find out what children know about reading before they have learned to read. 60 children (thirty boys and thirty girl) were interviewed about the subject and the results were strikingly similar between the groups: 95% knew letters were used to read and were different from numbers and drawings but only 10% said that letters "say" something. In trying to account for this 10%, we investigated their family environments.The results show that none of the factor investigated correlated with their metacognitive knowledge.

Nelson/Denman Rooms, 9:00 - 10:00 A.M.

This presentation examines how texts are talked into being in classrooms and how these texts form the basis for defining what counts as disciplinary knowledge in classrooms. Drawing of recent work in classroom ethnography, sociolinguistics, ethnomethodology, critical linguistics and New Literacy Studies, I will present an argument about the situated nature of disciplinary knowledge, of access to academic literacies, and of identity in classrooms. Through this work, I raise questions about theory-method relationships, units of analysis, and the claims that are possible using an ethnographically oriented approach to the study of text and discourse in classrooms and other institutional settings.

Denman Room, 10:30 A.M. - 12:10 P.M.

A central task of reading is to integrate currently-read information with information from earlier in the passage. The integration is often signaled by anaphors. Four experiments are presented in which several text variables (e.g., distance, a same category distractor) prevent noun phrase anaphors from being resolved. The results support a memory-based text-processing view; further, they are inconsistent with the view that anaphoric inferences initiate an exhaustive search of memory.

Three experiments tested whether or not instrument inferences are drawn on-line and before reading instrument identity in single sentences. Results suggested that instruments are generated predictively and on- line, contradicting previous research by Dosher and Corbett (1982) and McKoon and Ratcliff (1981). These contradictory findings are explained in terms of differences between verbs in their strength of implication of instruments. The results are discussed in relation to verb lexical entries and mental model theory.

In this study, we investigated the activation for predictive inferences as a function of the semantics features of the verbs expressing the consequence of the action. Using an on-line verification task, we observed that predictive inferences are higher activated in the texts describing intentional actions of a causal human agent vs  texts about changings produced by the action of forces of nature. This suggests that agent and object differed in salience in the situation model.

This paper investigates whether a specific type of inference, thematic inference, is generated on-line. A thematic inference is an inference about the global message of a text. The constructionist theory of inference generation predicts that these inferences are made in considerate text. The results of sentence reading times and word recognition judgements indicate that readers are able to generate thematic inferences off-line, but were inconclusive about the on-line status.

Nelson Room, 10:30 A.M. - 12:10 P.M.

Using the mental model approach to comprehension, this research evaluates the interaction of expertise and purpose for reading in computer programs. Expert programmers were expected to form a stronger situation model than novices. Participants given a read-to-do purpose for reading were expected to form a stronger situation model than those given a read-to-recall purpose. The results show an interaction between expertise and purpose for reading with novices building a stronger situation model if given a task requiring situation knowledge.

We investigated 11th graders' ability to integrate information when learning from multiple texts. We manipulated whether a story was told from the perspective of a single author or two different authors and the type of instructions provided. Students required to integrate information across two texts performed as well as those who read the story as a single text, but only when explicitly instructed on how to integrate. Students who read two texts with only comprehension instructions made more errors, recalled less information, and provided less integrated answers to questions.

The Vicarious Learner project is investigating the role of dialogue in learning and the benefits of 'overhearing' other learners. Our results suggest that dialogue serves to 'enculturate' learners into the patterns of language in particular disciplines and that such learning can occur not only through direct participation in dialogue, but also vicariously. We see educational dialogue, as opposed to everyday conversation, as being about the complex alignment of concepts where the participants know that an initial misalignment is fairly certain.

Over the century, researchers have been intrigued with similarities and differences between listening and reading. Most of the research has been based on a theory of a unitary nature of language processing using only a few sentences at a time, not extended discourse. In contrast, Horowitz & Samuels, 1985 demonstrated that poor readers out-perform good readers on listening to easy discourse but not in reading easy or hard discourse, suggesting discourse variation was an important factor in comprehension and that researchers could not generalize across modalities. The present study was conducted to look at oral-based versus literate-based discourse comprehension on two different topics. The oral-based versions were ghostwritten by a professional writer for delivery as speeches, while the literate-based versions were ghostwritten by the same writer for magazine articles (using language features identified as spoken or written forms, Biber, 1988). College students from two universities, and high school students, were administered tests of listening and reading. Results showed that oral-based texts were easier to process than literate-based texts, and that reading was more effective than listening, though it requires greater investment of mental effort. For some of the outcome variables, these factors interacted with the topics of the text. Research conducted supported a dual theory of discourse processing--based on discourse variability and modality of comprehension.

Denman Room, 1:50 - 3:30 P.M.

Researchers have long noted a relationship between informativeness and the pronunciation of words (Jespersen 1923): Less informative words are more likely to be reduced than more informative words (Jurafsky et al. 1998, Fowler et al. 1987, Resnik 1993). In this study, we investigate the effects on durational shortening of three possible sources of informativeness; the probability of a word given the two previous words, the number of times a word is used in the prior context, and the semantic word association between the word and the discourse context (as measured by LSA). From 789 examples from the transcribed Switchboard corpus of telephone conversations (Greenberg et al. 1996), we conclude that all three of these factors independently affect a word's informativeness in conversation, and hence its degree of durational shortening. Thus, we provide a model of the knowledge that speakers exploit during conversation.

Goal and source thematic roles have been shown to influence pronoun resolution, an effect that has been linked to the reader's tendency to focus on the consequences of the event (Stevenson et al., 1994). Using a story-continuation experiment and a corpus analysis, I show that speakers also tend to use pronouns more often for goal entities. Furthermore, speakers tend to refer more frequently to goal entities than source entities, a pattern which may explain the salience of goals.

We examined the effects of frequency (low, middle, high) and context on the interpretation of twenty frequency adverbs. For context-free adverbs, the variance in frequency ratings was highest for middle frequency, resulting in a curvilinear relationship between means and standard deviation of the ratings. Context only selectively decreased the variance in interpretations for adverbs of lower middle frequency (e.g., occasionally). Variance increased when the adverb modified an activity with a clear frequency norm.

Nelson Room, 1:50 - 3:30 P.M.

Evidence will be provided that demonstrates Autotutor's effectiveness as a tutor and conversational partner. We analyzed several tutorial transcripts in which Autotutor and students of varying ability levels collaboratively answered questions and solved problems. Knowledgable judges rated tutor dialogue moves on two dimensions: pedagogical quality and conversational appropriateness. Data from three evaluative cycles will be presented

The hinting strategy is one of the most powerful tools a tutor can use to facilitate a student's active knowledge construction within the zone of proximal development (ZPD). We reanalyze data on hints in a sample of human tutoring transcripts and present some new data from our current simulated tutor (AutoTutor). We will discuss ideal strategies of improving hints so they are more effective and engaging.

This research was concerned with two issues: the extent to which overhearers could both learn from listening to monologues and dialogues, and be induced to ask questions by listening to a large number of questions in a virtual tutoring session. In retention tests subjects in the dialogue condition wrote significantly more content than in the monologue condition. In a transfer task subjects in the dialogue condition asked significantly more questions than in the monologue condition.

Nelson/Denman Rooms, 4:00 - 5:00 P.M.

A new look at literal meaning in understanding what is said and implicated

What role does literal meaning play in language comprehension? This question has been vigorously debated in research on figurative language understanding. The "standard pragmatic view" proposes that people must analyze the complete literal meaning of indirect and figurative utterances before pragmatic information is consulted to infer speakers' nonliteral messages. Most of the psycholinguistic research shows, however, that given sufficient context people understand nonliteral meanings without first analyzing the complete literal meaning of an expression (i.e., the "direct access view"). Several lines of research have recently attempted to demonstrate that people still analyze aspects of literal meaning when understanding metaphors, irony, idioms, and proverbs. I critically evaluated this new work and suggest that it does not contribute sufficient evidence against the direct access view. Nonetheless, I argue that other research suggests how people analyze aspects of what speakers say as part of inferring what speakers implicate. This conclusion has several implications for specifying the role of pragmatics in ordinary utterance interpretation.

Sherrie Atwood
Simon Fraser University
P.O Box 91113
West Vancouver B.C. V7V 3N3
CANADA
atwood6@ibm.net

Alan Bell
University of Colorado at Boulder
Department of Linguistics
Campus Box 295
Boulder, Colorado 80309

Ricardo Benítez
Av. Brasil 2830
Edif. Gimpert
9º piso, of. 9-21
Valparais
CHILE
rbenitez@ucv.cl

Ronan S. Bernas
Department of Psychology
Eastern Illinois University
600 Lincoln Ave.
Charleston, IL 61920
cfrsb@eiu.edu

Nathalie Blanc
Institute of Psychology
University of Lyon 2 Laboratory of
Cognitive Psychology (E.A.D. Cognition-Emotion)
5 avenue Pierre Mendès-France
69676 Bron Cedex
FRANCE
nathalie.blanc@univ-lyon2.fr

Jonathan E. Bloom
New School for Social Research
Department of Psychology, AL-340
65 Fifth Avenue
New York, NY 10003
jbloom@newschool.edu

Marisa Bortolussi
Dept. of Comparative Literature, Film and Media Studies, and Religion
University of Alberta
Edmonton, AB
Canada T6G 2E2
marisa.bortolussi@ualberta.ca

Brian F. Bowdle
Indiana University
Department of Psychology
1101 East 10th Street
Bloomington, IN 47405
bbowdle@indiana.edu

M. Anne Britt
University of Pittsburgh
90 Fayerweather St.
Cambridge, MA
abritt@worldnet.att.net

Jean-Marie Burkhardt
Laboratoire d'Ergonomie Informatique
Universite Rene Descartes - Paris V
45, rue des Saints-Peres
Paris
FRANCE
Jean-Marie.Burkhardt@ergo-info.univ-paris5.fr

Dr. Jeremy Carpendale
Psychology Dept.
Simon Fraser University
8888 University Drive
Burnaby, BC V5A 1S6
jcarpend@sfu.ca

Miao-Hsia Chang
9F-1, #23, Alley 20
Lane 142 Sec. 6
Roosevelt Road
Taipei 117
TAIWAN
kc1010cc@ms10.hinet.net

Herbert L Colston
Psychology Department
University of Wisconsin-Parkside
Kenosha, WI 53141-2000
herbert.colston@uwp.edu

Frederick G. Conrad
Bureau of Labor Statistics
2 Massachusetts Ave., N.E., Rm. 4915
Washington, DC 20212
conrad_f@bls.gov

Richard Cox
Human Communication Research Centre
University of Edinburgh
2 Buccleuch Place
Edinburgh EH8 9LW
rcox@cogsci.ed.ac

Scotty D. Craig
The University of Memphis
Campus Box 526400
Memphis, TN 38152-6400
scraig@memphis.edu

Francoise Detienne
Action EIFFEL Bat 24
Domaine de Voluceau,
Rocquencourt, BP 105
78153, Le Chesnay, France
francoise.detienne@inria.fr

Finbar Dineen
Centre for Learning and Teaching Innovation
Glasgow Caledonian University
Glasgow G1 2RN

Rachel DiPaolo
University of Memphis
3226 South Mendenall #2
Memphis, TN 38115
raydip@thoughtware.com

Peter Dixon
Dept. of Psychology
University of Alberta
Edmonton, AB
Canada T6G 2E9
peter.dixon@ualberta.ca

Cynthia Doyle
Department of Psychology
The University of Memphis
CAMPUS BOX 526400
Memphis, TN 38152-6400
hgun@bellsouth.net

Miriam Eckert
Institute for Research in Cognitive Science
3401 Walnut Street, Suite 400A
University of Pennsylvania
Philadelphia, PA-19103
miriame@linc.cis.upenn.edu

Charles R. Fletcher
Department of Psychology
University of Minnesota
75 East River Road
Minneapolis MN 55455-0344
fletc002@tc.umn.edu

Christelle Font
University of Lyon 2, France
Hartwell S. Francis
hartwell.francis@colorado.edu