Eleventh Annual Meeting of the

The Eleventh Annual Meeting of

The Society for Text and Discourse

The University of California, Santa Barbara

Abstracts of the Presentations

Friday, July 13^th

10:15 - 12:45 Paper Session 1A

The time course of salience and context effects
Orna Peleg, Rachel Giora, (Tel Aviv University ) & Ofer Fein (The Academic College of Tel Aviv Yaffo)

Results from three experiments support the graded salience hypothesis, which assumes that initial comprehension involves two separate mechanisms (linguistic and contextual) that run independently. Experiment 1 shows that when the target word is placed in sentence-final position, contextual facilitation can occur even before lexical accessing is allowed. Experiment 2 shows that contextual effects do not outweigh salience effects at sentence-initial position, even when prior context is strong. Experiment 3 shows that when salience differences are polarized, contextual effects cannot inhibit salience effects even at sentence-final position. The role of salience in aesthetic creativity
Rachel Giora, Ann Kronrod, Idit Elnatan, (Tel Aviv University) & Ofer Fein (The Academic College of Tel Aviv Yaffo) Three experiments confirm that salience (familiarity, frequency, conventionality, prototypicality) plays a crucial role in aesthetic creativity. We argue that aesthetic creativity requires an optimal change such that (a) incurs a change of meaning, but (b) one that would also allow for the recovery of the salient meaning from which the novel meaning stems. Experiment 1 and 2 show that it is the expression that occupies a mid position on the familiarity scale that is most pleasurable. Experiment 3 shows that the pleasurable variation involves processing the familiar expression. Kaleidoscope: A comparative approach to poetic translation
Grace Po-ting Fang (University of East Anglia) My research topic is Kaleidoscope: A Comparative Approach to Poetic Translation. This project attempts to tackle the problems of presenting Chinese poetry in English translation through the form of hard copy and multimedia. My main concern is to find ways of representing the source text without any loss of meaning or effect and at the same time to achieve acceptability and to activate reader response in the target linguistic context. Interpretations of metaphor and simile in poetry
Zazie Todd (University of Leicester ) & David D. Clarke (University of Nottingham) This paper explores differences in the interpretation of poetic simile and metaphor. Forty participants completed a questionnaire responding to extracts from Neruda's poems, containing a phrase that was either a simile or (re-written as) a metaphor. Results showed a greater diversity of responses to metaphor than simile. The results also show the importance of subjectivity in interpretation of figurative language. The implication is that theories based on idealized models may not reflect actual readers' responses. The representation of characters' emotional responses: Do readers infer specific emotions?
Pascal Gygax, Jane Oakhill, & Alan Garnham (University of Sussex) The two experiments that will be presented show that readers do not, as previously assumed (Gernsbacher et al., 1992; Gernsbacher and Robertson, 1992; Gernsbacher et al., 1998; DeVega et al., 1996; DeVega et al., 1997), infer specific emotions while reading, but instead, infer only general emotional information that is shared by several more specific emotions. Results from Experiment 1 (off-line) show that participants judged several emotions consistent with the same story. In Experiment 2 (on-line), participants took longer to read target sentences containing emotions mismatching the stories, but there was no difference between target sentences containing different matching emotions. Cohesion in the coherence process: Evidence for the specification hypothesis
Max Louwerse (University of Memphis)
* Jason Albrecht Outstanding Young Scientist Award Winner In a series of experiments evidence is given for the specification hypothesis, which states that more specific cohesion relations facilitate the construction of a coherent mental representation. Recall is best for those situations that have the highest number of cohesion relations. Reading time shows robust specificity effects for cohesion relations, with those relations that are less specific requiring an incremental reading, while those that are more specific being processed in a wrapping up stage.

10:15 - 12:45 Paper Session 1B

Framing in computer-mediated communication
Kyong-Sook SONG (Dong-eui University, South Korea)

The concept of framing has influenced thinking about language in interaction in that no communicative move could be understood without reference to a meta-communicative message about what frame of interpretation applies to the move (Goffman 1974, 1981. Tannen 1993, etc). Based on various Internet Relay Chat(IRC) interactions, this paper investigates Korean speakers' management of frames with focus on their interactive, psychological, and communicative motivations. Framing is found a discourse strategy in Korean CMC, and interactive frames are better understood with reference to the relationship between language and culture. Online interaction: A discourse analysis of a graduate level webcourse
Judith C. Lapadat (University of Northern British Columbia) Although much has been written about the promise of online learning environments for higher education, few studies examine of nature of the discourse in such courses. This paper presents the results of a discourse analysis of an interactive, text-based, online, graduate education course, designed and taught according to constructivist principles. I will describe discursive characteristics and some devices used by participants in this online course, and provide examples of themes arising in the course that reveal processes of conceptual scaffolding and social negotiation of meaning. Expertise in delivering expert knowledge to laypersons: When do computer experts tailor their explanations to their listeners' needs?
Rainer Bromme (University of Münster, Germany) & Matthias Nueckles (University of Freiburg, Germany) Our study focuses on how information about the addressee and about the topic to be explained are used when explanations for laypersons are formulated. The experiment simulated an e-mail hotline scenario with computer experts as subjects. They had to answer fictitious e-mail queries from two different clients who varied with respect to their prior knowledge. Additionally the concepts which could be used for the explanation of a certain topic varied with respect their conceptual importance. Discourse in chair and couch psychoanalytic sessions
Catherine DiNardo, Michael F. Schober, & Jennifer Stuart (New School for Social Research) We compare form and content of discourse in an audio corpus of psychoanalytic treatments, where patients first faced analysts and later reclined on a couch with analysts sitting out of view. Although some psychoanalytic theories suggest the couch changes patientsí mental functioning and behaviors, we found few differences from chair to couch. We propose that participantsí perceived roles and goals create the patterns of interaction, overriding the odd visual copresence that using the couch creates. Discourse signals exchanged in supervision activities: Promoting or inhibiting the reflective thinking?
Andrea Claudia Ferreira Valente (Universidade Federal do Rio de Janeiro) This presentation is part of a case study on a novice teacher 's pedagogical reading practice at a language institute at the Federal University of Rio de Janeiro. Here, I examined the talk occurred during supervision activities where the researcher/supervisor attempted to interact dialogically with the teacher so as to stimulate reflection on her teaching practice. This study aimed at evaluating to what extent the supervision sessions helped the teacher reflect upon her pedagogical practice. This item is a grammatical one then?
Maria Alice Tavares (Santa Catarina Federal University, Brazil) Based on analyses of syntactic, semantic and pragmatic properties of the linguistic item então, I discuss the status of three uses of this item: inference introduction, pause filling and interjection. I also seek for evidences of what change process is involved in the development of the uses of então under analysis: grammaticalization or discursivization. I make use of real and present speech data from VARSUL (Urban Linguistic Variation in Brazil South Region) Project DataBase.

3:00 - 5:30 Paper Session 2A

Epistemological evaluations in the comprehension of expository text
Tobias Richter (University of Cologne, Germany)
* Outstanding Student Paper Award Winner

This paper investigates the role of epistemological evaluations (i. e., knowledge-based evaluations of text information) in the comprehension of expository texts. Comprehension involving epistemological evaluations should differ from cumulative knowledge acquisition, and may lead to 'knowledge with a point of view'. In an experiment, reading goals, plausibility, and learner characteristics were varied. Results of reading times and several offline indicators suggest that epistemological evaluations during reading were indeed associated with an epistemologically qualified text representation. Promoting understanding from electronic text: Windows, images and overviews
Jennifer Wiley (University of Illinois at Chicago) How can multiple-window environments support better understanding from text? In a series of experiments, students perform best when they contrast and compare information across texts, and when they garner evidence to support causal arguments. Using learning outcomes, written essays, and eyetracking measures as converging evidence, the present experiments examine which presentation format, including the number of browser windows, and the placement of images and overviews, supports the best comprehension and the most conceptual learning from text in multiple-window web-page environments. Structural representations in the comprehension of expository text
Hervé Potelle & Jean-François Rouet (Université de Poitiers and CNRS) The Landscape Model was used to generate a set of concepts that comprised the nodes in a structural representation of an expository text about the international oil market. Behavioral recall data (frequency of inclusion of gist concepts) was significantly correlated with the nodes generated by the Landscape Model. These nodes are being used in the design of structural representations that manipulate the placement of these concepts and impact of placement on comprehension and recall. Eyetracking and deep comprehension of illustrated texts
Elisa Cooper, Shannon N. Whitten, & Arthur C. Graesser (The University of Memphis) How do we know that a reader has obtained deep comprehension of a device? Eyetracking may be the answer. In this study, participants were asked to examine a situation in which a device broke down (breakdown scenario) and to generate questions about the scenario as we tracked their eye movements. Results showed that deep comprehenders spent more time fixating on a fault and fixated on the fault before generating their good question. Overhearing monologues and dialogues: Effects of vicarious learning on recall and question generation
Scotty D. Craig, Barry Gholson, Matthew Ventura, & David Driscoll (University of Memphis) Using a within subjects design and a vicarious-learning paradigm, participants heard information presented to them in both a monologue and dialogue format and where then tested on the content. Participants were then given an interactive session when they interacted with the experimenter on a different topic. Participants wrote significantly more content on questions that involved dialogue content. Furthermore, participants wrote significantly more after interacting with the experimenter than after initial acquisition. Construction of intertextual models in a 9th grade classroom
Susan R. Goldman & David Bloome (Vanderbilt University) In this presentation, we analyze a teacher-led discussion in a 9th grade language arts classroom that exemplifies socially-mediated sense-making. Sense-making in this instructional conversation involved constructing a meaningful representation of a text as well as a meaningful interpretation of the text. Interpretation involved integrating the text being read with prior knowledge including personal experiences, knowledge of prior texts, and knowledge of genres. The cognitive, social, and linguistic processes involved in this instructional conversation are highlighted to reveal one model of intertextuality for building literary meaning.

3:00 - 5:30 Paper Session 2B

Memory for lengthy discourse due to accompanying planned movement
Helga Noice (Elmhurst College ) & Tony Noice (Indiana State University)

This experiment investigated the memory benefits of movements that accompany lengthy discourse (a theatrical monologue). Professional actors were randomly assigned to one of three conditions. In the first, movements were planned and then physically executed while the participant verbalized the material; in the second, movements were planned and then only imagined while the participant verbalized the material; in the third, the material was deliberately memorized. Results showed that planning and imagining movement was sufficient to create enhanced memory compared to deliberate memorization. The genre diversity of dialogue game theory
William C. Mann (SIL International) Dialogue Game Theory (DGT) describes large-scale intention and action structures of dialogues. As part of an exploration of how and when Dialogue Game Theory fits well, this paper first identifies DGT, then examines several natural dialogues from diverse sources, and then identifies characteristics that they share. The dialogues are drawn from publicly available corpora, including London-Lund, LRDC Circle Archive, and the SRI/American Express Corpus. Non-dynamic discourse representation structure
Richard Breheny (University of Cambridge) The background for this presentation is the general re-appraisal of the dynamic semantic paradigm whose prototypical tenets are that linguistic meaning is context update potential and that utterance interpretation consists in the transformation of a commonly held discourse representation. In particular, in the light of Stalnakerís (1998) deconstruction of the dynamic analysis of indefinites and anaphora, we will focus on this key case for dynamic treatment. In this paper I will develop the truism that the pronoun condition makes reference to a very specific fact about interlocutors, that they are jointly attending something. I will make some proposals (based partly on early developmental literature - such as Tomasello 1995 - and partly on psycholinguistic research - such as Sanford and Garrod 1998) about the nature of joint attention and certain of its structural properties. Genre and lexicon: From index to speech act verb of clarification
Linda R. Waugh (University of Arizona) Genre provides a frame for lexical items: the semantic fields they occupy, their prototypical usage, and their patterns of contextualized senses. In this corpus-based study, I show that the French verb indiquer is a speech-act verb of clarification in journalistic usage. As such, through metaphoric transfer it has evolved away from its original visual indexical meaning but has still kept certain aspects of its original contextualized senses. Thinking for speaking in Basque: Evidence from oral and written narratives
Iraide Ibarretxe-Antuñano (University of California at Berkeley) This paper analyses the structure of motion events in Basque oral and written narratives. It takes up two related issues: First, it focuses on those specific areas where Basque presents particular characteristics when encoding motion and that are important for the ëthinking for speakingí in this language. Second, it draws a contrastive comparison between the strategies for coding motion events found in the two types of narratives. Data come from videotaped narratives of adult native speakers of Basque from different dialectal areas, and from three Basque novels. The production of overlap in agreeing and disagreeing responses
Carmen Santamaría-García (University of Alcala, Spain) Several studies claim that agreement is frequently produced in slight overlap with prior turns (Pomerantz 1975, 1984, Mori 1999). Disagreement, in contrast, is usually prefaced by pauses and hesitations. However, data from CSAE and Spanish conversations from CORLEC corpus suggest that overlap occurs more frequently than expected in disagreeing turns. A detailed study of overlap in agreeing and disagreeing responses is carried out to test this hypothesis.

8:00-10:00 pm Poster Session and Reception

P-1 Public discourse as the mirror of ideological change: A keyword study of editorials in People's Daily
Karen Wu Rongquan (City University of Hong Kong)

This study introduces empirical approach of keyword-context-meaning analysis to Chinese media discourse, an area as yet under explored, to interpret the changing content of ideology that China has experienced during the last 50 years. It examines the use of some 20 ideological keywords drawn from political reports of China Communist Party Congress in their practical discursive environment, i.e. the editorials from Peopleís Daily. The findings indicate that a) by identifying the collocations of keywords in different periods of history, one can trace the meaning change occurring in this word over years; b) by referring to the elements of ideology (which are defined by the concept we adopted in this study), one can interpret the changing content of ideology through a semantic study of keywords. P-2 Rhetorical styles in the First Congress
Juhani Rudanko (University of Tampere, Finland) The paper examines rhetorical styles of argumentation in three debates in the first United States Congress in 1789. The three debates dealt with procedural objections to considering James Madison's proposal for amendments to the United States Constitution. Different styles of argumentation are identified, based on specific argumentative devices. The relation of argumentative styles to external factors, including party affiliation, is also discussed. P-3 Room for further research in article introductions
Solange Aranha (Fundação de Ensino "Eurípides Soares da Rocha", Brazil) The purpose of this poster is to present the results of the use of CARS (Swales, 1990) in some introductions and also some evidences that justify a proposal of enlargement of the CARS model. We applied the model to 25 introductions from well-known journals and written by the editors or the members of the publishing board. We can observe most of the expected features in the introductions, although we believe an enlargement of the proposal should be considered in Move 3. P-4 An empirical investigation of live ink
Brian Linzie & Charles R. Fletcher (University of Minnesota) Live Ink is computer software developed by Walker Reading Technologies of Rochester, Minnesota. It is designed to format electronic text in a manner that facilitates comprehension by reintroducing some of the prosodic information that is normally lost when spoken language is written down. The research presented here shows that Live Ink has little impact on the comprehension of relatively easy narrative texts but produces marked improvements in the comprehension of more challenging expository texts. P-5 Causal connectivity and conceptual overlap in text processing
Joseph P. Magliano (Northern Illinois University), Michael B. W. Wolfe (Grand Valley State University), & Benjamin Larsen (Northern Illinois University) This study investigated the contribution of causal relatedness and conceptual overlap on online processing and memory for short texts. The contributions of causal processing should reflect conscious, explanatory reasoning, whereas conceptual overlap should reflect the automatic activation of information. The degree of causal connectivity between two sentences impacted reading time, whereas the degree of conceptual overlap did not. On the other hand, both factors had an impact on how related sentences are in memory. P-6 Predictive inference generation as a function of working memory capacity and causal text constraints
Tracy Linderholm (University of Florida) Two circumstances that affect the predictive inferential process are examined: working memory (WM) capacity and causal text constraints. A naming task was used in Experiment 1 and a reading time task in Experiment 2 to investigate predictive inferences based on low, moderate, and high causal sufficiency text events. The results of both experiments indicate that only high WM capacity readers make predictive inferences but do so based on highly sufficient text events. Thus only high WM capacity readers attend to causal text constraints when making predictive inferences. P-7 Facilitating effect of macro-structure intervening task during multiple texts integration
M. Anne Britt (Northern Illinois University ) & Jean-François Rouet (University of Poitiers, France) We investigated the effect that a macro-structure focusing task has on facilitating integration of information among multiple texts. In 3 experiments, we manipulated the task that was provided to participants between readings of two related texts. We found that on intervening task that focused the reader on the macro-structure representation of the initial text led to a more integrated final representation than either no task or a task that focused the reader on micro-structure or unrelated information. We are presently using van den Broek's Landscape Model (van den Broek, Risden, Fletcher, & Thurlow, 1996) to examine the time-course of inter-text activations. P-8 Production strategies for the construction of expository texts: A comparison between deaf and hearing students
Barbara Arfé (University of Padova, Italy) The production strategies of 8 prelingually deaf and 8 hearing students (from 3rd to 8th grade) in expository texts construction were analyzed using a technique of procedural facilitation and the relationship between students' production strategies and morpho-syntactical competence in writing was investigated. Results show that deaf and hearing participants used the same range of operations in text production, although a different effect of school grade level on students' strategies was revealed for the two groups. P-9 Reading and writing about science
Robert Calfee, Roxanne Miller, Evelyn Haralson, Crystal Howard, & Kathleen Wasserman (University of California, Riverside) This National Science Foundation three-year project aims to find out how to help teachers and students to master the research report, critical to the scientific enterprise. A pilot study was conducted in Summer 2000 and showed substantial increases in student writing (grades 4-8) in length, coherence, spelling, and vocabulary. The first full year of the experimental phase is now underway in Riverside and San Bernardino Counties of California. P-10 The impact of narrative structure on the memory for scientific text
Hyun-Jeong Joyce Kim & Keith Millis (Northern Illinois University) We tested the difference between narrative- and expository textbook organization on comprehension and retention. After reading either short narrative or expository chapters on psychology topics, participants were given short-answer and multiple-choice tests, and the Learning Style Questionnaire immediately or after one week. The narrative condition resulted in higher performance on the multiple-choice test than the expository condition for "low reflectors" when tested immediately, but the effect disappeared after a week. This suggests that narrative structures might have a short-lived advantage, but other results indicated that narratives increased interference. P-11 Writing & subjectivity: An analysis using systemic functional grammar of written texts from 3rd-12th grades of urban students
Judith V. Diamondstone (Rutgers, the State University of New Jersey) This poster presents a Systemic Functional grammatical analysis of the emergent styles of urban student writers responding to the same prompts given at different grade levels, from grades 3-11. Open-ended and rhetorically ambiguous, prompted writing invites stories built around the writers' unconstrained associations with the prompts. Extended interviews with the writers years afterwards provide a backdrop for the text analyses. P-12 Context, imagery, and abstractness
Xu Xu, Katja Wiemer-Hastings, &Jan Krug (Northern Illinois University) We examined factors underlying the abstractness of entities. Ratings were collected for two predictors, imagery and context availability. Consistent with earlier findings, we found that neither predicts the abstractness variance of abstract entities. We compared this to predictors derived from the contextual constraint theory: number and abstractness of contextual constraints. Test items were coded for constraints. The proportion of abstract constraints was a good predictor of abstractness. Theoretical implications are discussed. P-13 Metonymy and text cohesion
Abdul Gabbar Al-Sharafi (Durham University, England) This paper argues that metonymy is not only a semantic phenomenon limited to the lexical level of language but it is also a cognitive phenomenon that contributes effectively towards the interpretation of text. Metonymy is defined in this paper as a process of "stand for" between forms, concepts and objects, hence as a semiotic process. Perceived as such metonymy plays a major role in the understanding of the formal and semantic ties that make a text stand as a unified whole. A metonymic model of text cohesion will be presented and the various cohesive devices proposed by Halliday and Hasan (1976) will be discussed as metonymic processes contributing to the text generation, organization and interpretation. P-14 A simplest systematics for the organization of user-device interaction
Ilkka Arminen (University of Helsinki, Finland) In this study, I will analyze the interaction between a mobile internet device (WAP phone) and the user. WAP phone use can be analyzed as being composed of iterative steps. The user's basic step consists of taking an action with respect to current state of the device. The user's step may either alter the state of the device or it may preserve the state. This simple systematics allows us to distinguish the reasoning and negotion about next action from the evaluations of outcomes of user's actions. A simplest systematics may provide a possible starting point for the analysis of user-device interaction, in particular for simple devices, like WAP phones. P-15 Cognition and deixis in conversational narrative discourse
Tomoko I. Sakita (Doshisha University, Japan) Stylistic choices in conversational narratives reflect speakers' recall and conceptualization of past experiences. In report of recalled episodes, subjective and objective construals of recalled events lead to the choice of direct and indirect reporting discourse styles. Our perception of personal deictic relations, narrators' self-identity, and psychological distance from recalled persons influence narrative tense forms. The paper documents these points with cognitive models and presents quantitative and qualitative analyses of discourse examples. P-16 Dialogue acts and the common ground: Identifying structure in interactive discourse
Ilana Mushin (Univeristy of Melbourne / La Trobe University), Lesley Stirling (Univeristy of Melbourne), Roger Wales (La Trobe University), & Janet Fletcher (Univeristy of Melbourne) This paper is concerned with the identification of units of discourse that reflect the 'joint action' features of interactive discourse. We use a corpus of task-oriented dialogues to empirically evaluate the signalling of common ground (eg. Clark 1996) as the basis for identifying minimal units of dialogue (Nakatani & Traum 1999). Our results have important consequences for the computational modelling of interactive discourse which must, ultimately, take into account its emergent and dynamic properties. P-17 Repetition in conversation
Ali Gravier, Ana Perez, & Natalie Person (Rhodes College) The purpose of this study was to examine repetition across three conversational genres, natural conversations, human-to-human tutoring conversations, and computer-to-human tutoring conversations. We systematically analyzed both verbatim and paraphrased repetition expressions in order to document the baseline rates of such expressions both within and between the turns of conversational participants. In addition, we classified the particular discourse function of a random subset of the repetition expressions that occurred in the three conversation types. P-18 Resolving the inner conflict: The structure of neurotic discourse
Andrej A. Kibrik (Institute of Linguistics, Russian Academy of Sciences), Vera I. Podlesskaya, Tatíjana M. Kal'kova (Russian State University for the Humanities), & Alla O. Litvinenko (Moscow State University) We look at night dream stories told by neurotic and normal children (in Russian). The purpose of this study is to compare the stories by normal and neurotic children. The initial hypothesis is that normal and neurotic children produce substantially divergent discourse structures. We identify differences in terms of story complexity and in terms of particular rhetorical relations employed, and suggest that discourse structure can shed light on the cognitive structures of the narratorsí minds. P-19 The function of discourse markers in different conversation genres
Erin VanCleve, Abbe Kohl, & Natalie Person (Rhodes College) This research looked at the functions of discourse markers (and, but, oh, so, or, because, well, now then, also, okay, you know, I mean) in natural conversation, human-to-human tutoring, and human-to-computer tutoring dialogues. These markers were outlined in Schiffrinís (1987) book, Discourse Markers. The discourse markers were extracted from conversation transcripts, and the frequencies and functions of the discourse markers were documented. The primary purpose of this research was to gain a better understanding of when and how particular discourse markers function in learning dialogs versus natural conversation. P-20 Cognitive basis of incongruity in verbal humor
Rachel G. Hull & Jyotsna Vaid (Texas A&M University) Incongruity effects in humor generation and perception were examined in two studies. Experiment 1 elicited humorous similarities between concepts (e.g., MONEY and CHOCOLATE). Experiment 2 elicited properties of the concepts presented individually, ranked by output dominance (frequency). It was hypothesized that humorous responses would more often include diverging output dominance scores of the constituent concepts whereas non-humorous responses would contain overlapping OD scores. The hypothesis was supported. Implications for models of humorous discourse are discussed. P-21 Understanding irony: On-line processing of figurative and literal meaning
Stacey Ivanko & Penny M. Pexman (University of Calgary) The purpose of the present research was to investigate the role of context in processing figurative and literal meanings of sarcastic-ironic statements. We manipulated degree of situational negativity for literal and sarcastic statements, and results showed that reading times for sarcastic statements were slower than for literal statements in the strongly negative situations. However, reading times for sarcastic statements were faster than or equivalent to reading times for literal statements in the weakly negative situations. P-22 Understanding positive and negative metaphor and irony: Off-line interpretation and on-line processing
Penny M. Pexman & Kara M. Olineck (University of Calgary) We investigated how ironic insults and ironic compliments are understood (Experiment 1, off-line ratings task) and processed (Experiment 2, on-line moving window task) and found that interpretation of ironic insults depended on whether participants judged speaker intent or listener perception. This may explain previous inconsistent findings regarding interpretation of ironic insults (Colston, 1997; Dews & Winner, 1995). We also found correlations between our off-line and on-line data, suggesting that on-line processing is related to off-line interpretation in particular ways. P-23 Using a causal network to model memory for an unscripted social interaction
Aaron Brownstein & Stephen J. Read (University of Southern California) We used Trabasso and van den Broekís model of memory for narrative text to predict memory for a videotaped social interaction, an episode of "Cheers." Participants watched the videotape and then recalled it. We constructed a causal network of the sequence; for each concept measuring whether it was on the main causal chain, how many links it had and total link strength. All three predictors significantly predicted recall, but only number of links predicted uniquely. P-24 Participation frameworks and interactional sequences in shared book reading between parents and children with SLI
Judy Vander Woude(Calvin College) & Ellen Barton (Wayne State University) In this paper, we identify and describe the participation frameworks and interactional sequences used during shared book reading between parents and children with specific language impairment (SLI). We show how interactional routines are combined in lengthy exchanges utilizing different types of questions, a design that positions the child as a communicatively competent participant in the interaction. We use this data to argue against a deficit-oriented model of language and literacy development in children with SLI.
Saturday, July 14^th

10:00 - 12:05 Symposium

Automated Techniques for the Analysis of Discourse

In this symposium, four talks will be presented which illustrate the range of domains to which automated discourse analysis has been applied. The talks will address both the methodological issues confronted in analyzing the discourse as well as describe the results and the implications of the studies. The talks will all present an overview of prior research in their domains as well as novel results from their analyses. The four domains of discourse that will be described will be the analyses of verbal protocols generated during reading (Magliano, Millis, Wiemer-Hastings, and McNamara), tutoring dialogues (Graesser, Person, Louwerse and the Tutoring Research Group), clinical interviews of schizophrenic patients (Elvevåg, Foltz, Weinberger and Goldberg), and communications among soldiers performing team tasks (Foltz, Cooke, Kiekel and Shope). Each talk uses common techniques but applies the techniques to widely different types of discourse . In addition, the talks address a number of research issues within cognitive psychology, including individual reading ability, group cognition, the pedagogical issues of interactions among learners and trainers, and the effects of mental disorders on cognitive processing and organization.

While the talks cover disparate areas, a common theme from the symposium will be the focus on introducing the methodologies used and the implications of using the methodologies. Each speaker will address as common set of questions:

What other techniques have been used in the past for dealing with this type of discourse?

What do these techniques allow us to do that we haven't been able to do in the past?

What are the advantages of the techniques for this type of discourse?

What are the limitations of the techniques for this type of discourse?

What are the theoretical implications to using this approach?

An introduction to LSA
Peter Foltz (New Mexico State University), symposium organizer

Tutorial dialog
Arthur C. Graesser (University of Memphis), Natalie Person (Rhodes College), Max Louwerse (University of Memphis) & the Tutoring Research Group

Using LSA to reveal reader strategies
Joe Magliano, Keith Millis, Katja Wiemer-Hastings, (Northern Illinois University) & Danielle McNamara (Old Dominion University)

Analysis of clinical interviews of patients with schizophrenia
B. Elvevåg (Clinical Brain Disorders Branch, NIMH/NIH), P. Foltz (New Mexico State University), D.R. Weinberger & T.E. Goldberg (Clinical Brain Disorders Branch, NIMH/NIH)

Automating measurement of team cognition through analysis of communication data
Peter W. Foltz, Nancy Cooke, Preston Kiekel, (New Mexico State University) & Steven Shope (Sandia Research Corp.)

10:00 - 12:05 Paper Session 3B

Discourse filler or topic marker
Dami Lee & Myung-Hee Kim (Hanyang University at Ansan, South Korea)

This paper examines 196 Korean narratives of three age groups and investigates the phenomenon of NP+-yo where young children tend to use politeness sentence ending ? yo inside the sentence after a noun phrase. Assuming that it serves some temporary discourse functions, we postulate two hypotheses: i) as a discourse filler due to processing constraints, and ii) a topic marker resulting from childrenÕ s limited discourse competence. All evidence considered, we cannot reject either of the hypotheses. An epistemic analysis of the Mandarin LE: An integrated approach
Kylie Hsu (California State University, Los Angeles) This paper examines the controversial temporal marker LE in Mandarin grammar. It proposes an epistemic approach to analyzing the underlying semantic substance of LE as "completion" and the various surface interpretations of LE in discourse. It allows a unified treatment of LE in linguistic analysis and explicates the seemingly disparate functions of LE. This novel approach also accounts for unusual cases of LE that are not analyzed in previous works. A contrastive study of discourse intonation systems of English and Japanese
Ken-Ichi Kadooka (RYUKOKU UNIVERSITY, Japan) The English intonation system is defined as Pragmatic Intonation System since it carries the speakerís expression of various dynamic aspects of interpersonal nuances such as irony, surprise, bargaining with the hearer(s). The Japanese counterpart is classified as Syntactic Intonation System because it is less expressive in pragmatic spheres than that of English, but demonstrative in distinguishing semantic and/or syntactic clauses of declaratives, interrogatives, or imperatives. Pronouns and their referents: Referent-tracking in Finnish
Elsi Kaiser (University of Pennsylvania) In this paper, I show that to successfully tackle referent tracking and pronoun resolution in a language such as Finnish, a free-word language without articles, we need to refine the "peg system" used in dynamic semantics. I provide an algorithm which uses the pragmatically-motivated word order tendencies of Finnish to create an ordered register of pegs (where each peg is associated with an entity in the discourse), ranked according to salience. Pronouns are interpreted as referring to the topmost (most salient) peg in the register. Shifting between frames of reference in Tongan linguistic descriptions of space
Giovanni Bennardo (Northern Illinois University) First, I introduce the concept of frame of reference and present a typology of frames. Then, I discuss the results of analysis conducted on data collected in Tonga, Polynesia. Linguistic tasks were administered in order to obtain language production in which spatial relationships between objects had to be described. The analyses show that Tongan use more the relative and the intrinsic frames of reference in small-scale space and the absolute frame of reference in large-scale space.

3:55 - 5:10 Paper Session 4A

Addressees' needs affect speakers' syntactic choices
Calion B. Lockridge & Susan E. Brennan (State University of New York at Stony Brook)

Brown and Dell's (1987) experiments suggested that apparent adjustments to (confederate) addressees' needs simply reflect what is easiest for speakers. We had speakers retell stories containing typical or atypical instruments to naive addressees. Speakers mentioned atypical instruments in the same clause as the action verb more often when addressees lacked an illustration than when they shared speakers' illustrations. This suggests that with visual copresence, speakers can take addressees' knowledge into account early in syntactic planning. Oral joke performance versus written joke text
Neal R. Norrick (Saarland University, Germany) My paper reports research comparing oral joke performances with written joke texts. I began with jokes from conversation and professional comedy performances, then asked subjects to retell jokes they had read. Oral joketelling adheres to characteristic conventions encouraging multiple processing. Joke recipients must listen for two sets of contradictory cues: (1) lexical and structural elements of the joke text geared toward obscuring the jocular interpretation and (2) prosody and timing elements geared toward revealing it. Prosody, good news, and bad news in conversation and literature
Raymond F. Person, Jr. (Ohio Northern University) Freese and Maynard described the prosodic characteristics of good and bad news in conversation. This paper analyzes how these features are represented in reports of good and bad news in literature. Literary discourse represents these prosodic features in various ways, including the use of punctuation and lexical items such as verbs and adverbs (e.g., "cried Jo passionately"). Thus, literary discourse presents these prosodic elements to readers, who co-produce meaning relating to good and bad news.

3:55 - 5:35 Paper Session 4B

News values and standpoints in the news
Luuk Lagerwerf (University of Twente, Netherlands)

In this paper, it will be shown how the specification of standpoints can be used to perform content analysis of latent content. From news value factors of some news event, standpoints can be derived. In a corpus of news items covering the event, argumentative markers are identified that support or deny one of the derived standpoints. The frequency of argumentative markers is a measure for the latent content being communicated. The difference between 'yes' and 'no'
Bregje Holleman (Utrecht University, Netherlands) Survey questions worded with the verb 'forbid' prove not to elicit opposite answers to equivalent answers worded with the verb 'allow'. Although 'forbid' and 'allow' are generally considered each other's counterparts, respondents rather answer 'no, not forbid' than 'yes, allow'. In order to find out which question is a more valid measure of the underlying attitude, this asymmetry in the answers has to be explained. The results of two correlational experiments show that the asymmetry arises because respondents translate similar attitudes differently into the answering options due to the use of forbid/allow. How does this translation process work? Results of 10 experiments investigating the meanings of the answering options to forbid/allow questions show how 'yes' and 'no' differ in their extremity and in the extent to which their meanings are well-defined, due to the use of 'forbid' or 'allow'. These semantic differences explain for the asymmetry in the answers, and build up to some general recommendations concerning questionnaire design. Coding unsegmented free response data using AUTOCODER
Richard M. Golden, Cindy Jaynes, Jason Earwood, & Michael A. Durbin (University of Texas at Dallas) If the process of coding protocol data could even be partially automated to increase both reliability and documentation effectiveness, this would be an important scientific advance. An HMM algorithm for automatic coding and segmentation of protocol data is described. For these simple texts, agreement measures for the test data (i.e., data not used to estimate the HMM algorithm's parameters) ranged from 80% to 85% for unsegmented text with the corresponding Kappa scores in the range of 55% to 69%. Investigating automaticity in L2 acquisition: Methodological issues
Olaf Bärenfänger (University of Bielefeld, Germany) & Maria José Peres Herhuth (University of Heidelberg, Germany) The goal of our paper is to identify typical features of automatized L2 oral speech production. After discussing essential properties of automatized cognitive processes in general we will turn to characteristics of automaticity in L2 acquisition. The oral output of a group of learners of German as a foreign language is finally compared to the predictions derived from our theory of L2 automaticity.