There is evidence that knowledge of the structure of a text facilitates comprehensibility and recall. In investigating both structure and content variables, Thorndyke (1977) found that
"Comprehensibility and recall were found to be a function of the amount of inherent plot structure in the story, independent of passage content. Recall probability of individual facts from passages depended on the structural centrality of the facts: Subjects tended to recall facts corresponding to high-level organizational story elements rather than lower-level details." (1977:77)
This fitted in with a great deal of other work on schema theory (Bartlett, 1932; Rumelhart, 1975; Schank & Abelson, 1977).
Kintsch, Mandel & Kozminsky (1977) used scrambled stories to investigate text macrostructure. They concluded:
"….the reader's story schema permits him to reorganize the scrambled story, if that story corresponds to his schema"
"…that the macrostructure of a story is formed during reading, as part of the comprehension process…" (1977:552)
Similar results were obtained in work by Mandler & Johnson (1977), and Kintsch & van Dijk (1975). Cirilo & Foss also found that "readers are sensitive to the structure of a story as they read it" (1980:104) and were able to distinguish high and low level propositions. See also Whaley, 1981.
All this work was conducted using narrative texts. Thorndyke proposed that "like sentences, narratives have their own internal structure" and grammar rules for simple stories were proposed. But similar results were found in other text types including expository texts. Meyer (1975a,b) found five basic organisations of discourse: collection, description, causation, problem/solution, and comparison.
Meyer and Freedle (1984) investigated the effects of different discourse types on memory. They found that the more organised types of discourse such as comparison and causation facilitated learning and memory.
Taylor (1980) found that children who were more sensitive to text structure recalled more expository material than those who were less able to use organisational structure of expository text. Likewise, Meyer, Brandt & Bluth (1981) investigating a reading strategy (identifying and using an author's prose organisation) found " a strong relationship between comprehension skills and the use of the top-level structure in text." (1981:82). McGee (1982), and Richgels, McGee, Lomax & Sheard (1987) came to similar conclusions.
All of these studies are concerned with high level organisation concerning whole stories or long stretches of expository prose. But it is clear that there are lower levels of organisation from the paragraph level down to relations between individual sentences and clauses. And, not surprisingly, clear structure on these levels also facilitates comprehension. Kieras investigated structure in simple paragraphs. He found that "paragraphs that violate the coherence and topicalization conventions yield longer reading times, poorer recall, and distortion of apparent theme" (1978:27)
All the above studies were conducted with native English speakers. Carrell investigated the effect of different prose organisations on the reading comprehension of ESL readers of various L1 backgrounds. She asked
"whether different groups of ESL readers possess the formal schemata against which to process these various rhetorical structures [and] whether there is a differential impact of these various rhetorical structures on different ESL readers." (Carrell, 1984:449)
She found that
"as [..] for native English readers, some variations in discourse type influence the amount of information recalled from prose by ESL readers. Further, the more highly organised types of discourse are generally more facilitative of recall than the less organised collection of descriptions." (Carrell, 1984:458)
Urquhart (1984) investigated the effect of rhetorical ordering (specifically, time-ordered and space ordered texts) on readability and, unsurprisingly, concluded that well ordered texts were easier to read.
Carrell also found that the effects of discourse type were not the same for each language group (1984:460) indicating that text readability is also a function of the text schemata that readers from different L1 backgrounds possess. (See also Connor & McCagg, 1983: Connor, 1984)
"by far the most interesting finding of this study is the result that if ESL readers possess the appropriate formal schema against which to process the discourse type of the text, and if they utilise that formal schema to organise their recall protocols, more information is retrieved." (Carrell, 1984:464,465)
It might be a strong claim to suggest that there are certain types of rhetorical organisation which are more readable (i.e. comprehensible, memorable) than others in any absolute sense, since there are so many other factors involved. Perhaps some are more readable simply because they match our expectations about how a narrative or expository text should be structured. Also, causation and comparison are perhaps, in general, more memorable because they require deeper levels of processing than descriptions. Whatever the case, since L2 learners are going to have to read more of these texts it seems only fair to make sure that the texts they are presented with conform to recognised types. As with all learning, it is difficult to induce patterns if we are presented either with aberrant patterns or with too few examples of each type of pattern. (See Evans, 1967 cited in Perkins & Angelis, 1985). One reason for using the same type of texts with learners rather than a selection of different texts often to be found in reading skills texts is so that students can familiarise themselves with these formal schemata. Hopping about from text type to text type gives the learner no chance to make inferences about any one type. Texts whose formal schemata do not conform to recognised structures should be classified as difficult (all other things being equal).
It may be possible to classify certain formal schemata as more or less difficult for particular readers. But this can only be done by those who are familiar with the difficulties those readers have.
Carrell (1992:16) cautions against the assumption that particular text types are always more difficult than others: "there appear to be complex relationships among particular types of text structures and given groups of subjects, and one cannot assume what the results would be if different types of text structures and/or different groups of subjects were investigated."