CHAPTER 2: LITERATURE REVIEW
2.2 L1 and L2 Discourse Processing and Reading Comprehension
2.2.3 The Construction Integration Model
Kintsch’s (1986, 1998) model of discourse processing provides a current and useful
framework for understanding the cognitive processes involved in several levels of reading comprehension (Horiba, 2000, 2002; Meyer Sterzik & Fraser, 2012; Nassaji, 2007; Perfetti & Stafura, 2014; Yoshida, 2012; Zwaan & Singer, 2003). Kintsch (1986, 1998) differentiates between understanding and learning from texts with three levels of mental representations: parsing, textbase, and situational. At the parsing level, readers are concerned with clauses and sentences which they understand; this is the process of decoding the text using only the linguistic features of the text and the reader’s knowledge of linguistic information (Kintsch, 1986). At the parsing level, no inferences are generated (Kintsch, 1998); it is a “shallow” representation with no “referential specification for each noun” (Graesser et al., 1994, p. 373). As inferences are not generated during parsing, this level
of comprehension is not relevant to the current study.
Comprehension, in Kintsch’s (1998) model has two main processes: construction and
integration. In construction, the reader decodes the letters and words into micro-propositions and propositions, activates prior knowledge, attaches each proposition to prior propositions, and mentally categorizes the text into a hierarchal structure. The textbase representation focuses on propositions and their organization into micro and macro structures within the reader’s memory.
The textbase level includes the application of the reader’s formal schemata of rhetorical structure to organize the propositions. The result is a textbase representation or understanding. If successful, the author’s intended meaning is understood, and Kintsch (1986, 1998) associates this comprehension
process with remembering a text. This textbase is a literal understanding of the text; it does not include analysis, assessment, nor application of the content to novel situations. The reader is able to
accurately report the content of a text in a summary type assignment, but does not go beyond the text.
Kintsch (1986, 1998) further contends that skilled readers are able to attain a deeper level of comprehension by integrating the content with their existing knowledge. This integration of the textbase with the reader’s knowledge is described by Kintsch (1998) as a situational representation
and is associated with text interpretation and learning. A situation model of a text is a mental representation of the situation described by a text (Kintsch, 1986). For example, if reading a text that describes a country, the situation model is the mental map, or picture, of the country that the reader creates. A textbase is required to construct a situation model of a text, but a textbase does not guarantee that a reader will create a situation model of a text (Kintsch, 1986, 1998). It is also
possible that an inaccurate textbase could be used to create a situation model, but the accuracy and quality of the situation model may, thus, be questionable.
Kintsch (1986) illustrates the differences between textbase and situation model representations with four experiments, one with children given easily and difficultly worded problems, and three with college students with survey and route style maps. The findings
demonstrated that there were no significant differences in the recall of easily and difficultly worded arithmetic problems for the children that did not solve the problems; however, those that solved the problems, or created a situation model of the word problems, remembered significantly more easy problems than difficult problems. Kintsch (1986, 1998) asserts that this is because the children were able to create a situation model because they understood how to use the linguistic information in the simpler worded problems to create a situation model; whereas, with the more difficultly worded problems, they were unsure how to solve them, so they were unable to create an effective situation
model. This demonstrates that having to solve a problem requires the reader to reconstruct the linguistic information into a mental situation, thereby, modifying the linguistic information which fosters retention (Kintsch, 1986).
College students showed similar results with textbase and situation models of text
representation (Kintsch, 1986). Participants were given one of two styles of text, and both styles of text were coherent, but one was more so; it was recalled at a significantly higher rate than the other. As has been demonstrated by much research, text coherence matters in constructing a textbase representation (Grabe, 2009; Kintsch, 1986; Koda, 2005). In regard to learning from a text, however, coherence did not have an effect, but the type of mental representation did. Recall output amounts were almost exactly the same for the two groups, but those that created situation models of the text by drawing a map were far more successful in their content recall (Kintsch, 1986). The textbase group largely reproduced the original text with only 24% of their output not matching the original text. Kintsch labeled the non-matching output as inferences and elaborations. In that portion of their recall, 79% of the inferences and elaborations made were wrong. The situational group reproduced fewer words from the original text, but their elaborations and inferences (35%) were correct 82% of the time. Kintsch (1986) concluded that reading to remember and reading to learn are based on different cognitive processes.
Kintsch’s (1998) model is of relevance to this study as his framework is used to distinguish
the levels of comprehension required for the tasks discussed in the relevant studies as well as the current study. However, a classification of how different reading purposes interact with cognitive processing is needed for the discussion of the literature because there are multiple research focuses and purposes, and not all studies have been framed with reading purpose specifically. Carver
(1997) provides a framework that can consistently be applied to many reading studies and, as described in the following section, proposed a theory describing the specific cognitive processes required for different reading purposes.