The Science of Reading. Группа авторов

Чтение книги онлайн.

Figure 1.3 The comprehension system of the Reading Systems Framework.

       From global top‐down structures to actual comprehension

      Text comprehension results from word‐by‐word, phrase‐by‐phrase, and sentence‐by‐sentence processes that are challenging to study. So, research started at the other end – where global structures could be seen as shaping local word and sentence processes. Early artificial intelligence (AI) systems started with global organizers for restricted situational comprehension (Schank & Abelson, 1977). Similarly, approaches within psychology and education also emphasized situated conceptual structures or schemata (Anderson & Pearson, 1984). Evidence for global top‐down guidance came from studies showing that a nearly incomprehensible text could be understood with a helpful title (Bransford & Johnson, 1972) and that a text lacking referential specificity could be understood as being about either music or card playing depending on whether the reader was a student in music education or physical education (Anderson et al., 1977).

      Other approaches focused on more generalized mental structures (e.g., story grammars, Mandler & Johnson, 1977; Stein & Glenn, 1979) that guide narrative comprehension. Trabasso and colleagues (1984) argued that people seek causality in reading stories and showed that causal expectations predict how readers understand sentences (Trabasso & Suh, 1993). In Reading Systems Framework terms, these approaches focus on the general knowledge component and largely ignore comprehension processes. They provide demonstrations of global influence without dealing with the nuts and bolts of comprehension.

       Text comprehension from the bottom up

Kintsch and van Dijk (1978) approached text comprehension as a cyclical process, with every text element activating meanings on the way to a coherent representation of the text as a whole. The Construction Integration model (Kintsch, 1988, Kintsch & Rawson, 2005) proposed two phases of comprehension: An initial construction phase, prompted by word meaning, spreads activation across memory of both text elements and general knowledge in a passive, automatic process.¹ A companion integration phase uses the overlap of meaning among the activated elements to constrain what information remains for the next cycle. Multiple integration phases lead to a coherent representation of the text.

The Construction Integration model moved text comprehension research toward a processing approach, incorporating memory‐based, word‐meaning, and sentence level components. The structure building framework (Gernsbacher, 1990, 1997) emphasized the complementary processes of memory‐based meaning mapping and structure building. Later models retained this focus on bottom‐up, memory‐based processes, including the Resonance Model (Myers & O’Brien, 1998) and the more recent Resonance, Integration, and Validation Model (Cook & O’Brien, 2014).

      Global influences continued to be emphasized in constructivist theories that assume readers are driven to construct coherence and search for meaning (Graesser et al., 1994). Top‐down influences were elaborated more specifically as mental structures to guide the reader’s construction of coherence, for example, dimensions of time, space, and causality in the Event‐Indexing Model (Zwaan et al., 1995). The Landscape model combined the automatic bottom‐up processes of memory‐based models with the top‐down influences of constructionist theories (van den Broek et al., 2005; van den Broek et al., 1999). In this model, a coherent mental representation emerges from both text and external knowledge activation patterns that increase and diminish over the course of reading a text. Comprehension results from the mixing of automatic passive processes with reader‐initiated strategic processes determined by the reader’s standard for coherence and goals in a particular reading situation (van den Broek et al., 1995; van den Broek & Helder, 2017; see van den Broek & Kendeou, this volume).

       The situation model: Knowledge and inferences

      Text comprehension results in memories at multiple levels, two at minimum: the text surface and the mental‐model (Johnson‐Laird, 1983). Research generally has followed the three‐way distinction of van Dijk and Kintsch (1983): surface level, text‐base, and situation model. This three‐way distinction adds a level of language‐based text meanings (propositions) intermediate between clauses/sentences and situated meanings.

      Critical in the situation model are inferences that require knowledge from both the text and the reader’s general knowledge. Bridging inferences are often required to make a text coherent (see O’Brien et al., 2015). For example, in reading “The bright sun lit the field. Alfred’s snowman melted,” one maintains coherence by inferring that the sun’s heat caused the snow to melt (Singer et al. 1992). When related knowledge triggers elaborative inferences, which are not required for coherence, comprehension becomes referentially richer and more interpretative, although unwarranted inferences can lead to inaccuracies. Successful comprehension СКАЧАТЬ