出版社：Centro Interamericano de Investigaciones Psicológicas y Ciencias Afines
摘要：El propósito de este trabajo consistió en es- tudiar el rol de la generación de inferencias en la comprensión de textos narrativos; para ello se re- alizaron dos simulaciones con el Programa Landscape sobre un texto narrativo natural, una simulación con arreglo de inferencias emocio- nales y otra sin ellas, que se pusieron en relación con datos conductuales de sujetos, obtenidos del reconocimiento de las oraciones del texto y de su valoración o relevancia para la historial. Con este propósito, participaron 30 adultos uni- versitarios. Leyeron la narración y posterior- mente respondieron a un protocolo de reconoci- miento y valoraciones de proposiciones del tex to. A continuación se llevó a cabo un análisis de correlaciones y luego un análisis de regresión lineal empleando los valores de proposiciones predichas a partir de dos simulaciones, una si- mulación causal-referencial (sin emociones) y otra con implementación de inferencias emocionales, como variables predictoras, y los va- lores obtenidos de reconocimiento y valoración y/o relevancia para la historia, como variables dependientes. El análisis de correlación mostró que ambas simulaciones se asociaron con los va- lores de reconocimiento y valoración, aunque en mayor medida, la simulación con implementa- ción de inferencias emocionales; pero el análisis de regresión detectó que únicamente la simula- ción con implementación de inferencias emo- cionales explicó los datos obtenidos de recono- cimiento y valoración. Estos hallazgos sugieren que las inferencias emocionales juegan un rol de importancia en la comprensión de textos narrativos, ya que per- miten que el lector focalice su atención hacia de- terminados puntos de la historia, que serán los nucleares o más importantes.
其他摘要：Textcomprehension requires the construction of a coherent mental representation, integrating text information with previous knowledge. Several studies have suggested that the emotional states of characters need not be stated explicitly: readers c an infer them as a consequence of the narrative situation, characters’ goals, actions, and relation s to other characters.This study investigated emotional inferences generation during reading using the simulation of a narrative text, carried out with th e Landscape Program and assessing text sentence recognition and the relevance of the sentence to th e story. This program is a connectionist model that represents comprehension as a changing landscape of activations of propositions along several readin g cycles. This model proposes that, as the reader procedes through a text, propositions fluctuate in activation. That is, with each reading cycle, new propositions are activated, and activation values o f current propositions change. In addition, the co- activation of propositions leads to the establishme nt of connections between them. Through these fluctuating activations, a memory representation of the text gradually and dynamically emerges. The peaks and valleys of this landscape represent the relative contribution of each proposition at any given point in the story, and are the base for the construction of a mental representation of the stor y. The aim of this work is to determine if compre - hension of a narrative with emotional inferences is a better predictor of textual proposition recogniti on and its relevance to the story than comprehension of a narrative without emotional inference, using a program that simulates the comprehension process. In order to simulate the generation of emotional inferences, we used the Landscape Computational Model. For this purpose 30 participants, unde r- graduates (9 males -30%- and 21 women, with a mean age of 20.67 years, SD = 2.85), read a story and subsequently completed a proposition re - cognition and proposition relevance for the story protocol. Pearson correlation analysis and linear regression analysis were conducted. Two linear regression models were tested, both including propositional values of simulation with emotional inference and without emotional inferences as independent variables, one including proposition recognition as dependent variable, and the othe including proposition relevance for the story as dependent variable. The Pearson correlation coef - ficients showed that the simulation of the story wi th emotional interference and the simulation of the story without emotional interference are related to proposition recognition and story relevance, al - though the relation between this values and the simulation based on emotional inference had better coefficients ( r = .35, p < .01, and r = .38, p < .01 respectively) than the causal-referential simulatio n ( r = .27, p < .05, and r = .28, p < .05 respectively). The linear regression analysis detected that only t he simulation with emotional inference explained the variance of recognition data (β= .45, p < .01) and the variance of the relevance to the story values (β= .53, p < .01). These findings suggest that emotional inferences play an important role in the understanding of narratives texts; because they focus the reader’s attention to certain important points of the story. That is, the realization of emotional inferences seems to intensify the attention that the reader devotes to the entire cyc le (which includes the proposition that prompts the inference, and also those that are causally connect ed to it), facilitating its later recognition and rele vance to the story. This intensified processing can be related to the role that characters’ emotional reactions play in a narrative.
