what the research says

Comprehension may not be as simple as the 'Simple View' suggests

Approximately two decades ago Gough and Tunmer (1986) proposed the Simple View of Reading (SVR) according to which reading comprehension (RC) is equal to the product of two separate components; decoding (D) and linguistic comprehension (C), thus RC=D x C

Georgiou, G.K., Das, J.P., Hayward, D. (2008). Revisiting the "Simple View of Reading" in a Group of Children with Poor Reading Comprehension. Journal of Learning Disabilities, 42 p.76-84.

Consistent with previous studies (e.g., Das et al., 2007; Janzen, 2000), the participants performed below grade level in reading comprehension (Grade Equivalent = 2.5, M[ean] standard score = 87.86, SD [standard deviation] = 8.29), but within grade level in decoding (Grade Equivalent= 3.9, M standard score = 98.16, SD = 8.29) and listening comprehension (Grade Equivalent = 3.9, M standard score = 100.62, SD = 7.99). If both decoding and listening comprehension are within grade level, then their product (D x C) should be within grade level. However, reading comprehension was well below grade level.

Georgiou, G.K., Das, J.P., Hayward, D. (2008). Revisiting the "Simple View of Reading" in a Group of Children with Poor Reading Comprehension. Journal of Learning Disabilities, 42 p.76-84.

The authors do not speculate on what, for these students, might be needed in addition to decoding and listening comprehension.

On teaching inference as a comprehension strategy

This is a little long, but gets to some important insights into inferences and how to teach comprehension:

Consider the following short section of a story:

Janet's dog Brutus slammed into the table in his enthusiasm. Later she surveyed the broken shards of her rare vase.

There are many inferences that could be made when reading this passage. First, inferences connect anaphors to their referents. Anaphors are those terms that refer to a character or object in a new way. For example, in the first sentence, readers infer that Brutus is the enthusiastic male referred to by the pronoun ‘his'.

A second category of inferences that are important to make while reading are those that are logically required for maintaining causal coherence; that is, establishing the causal relations among events.In this short passage, readers must infer that the vase was on the table that was bumped, and that the vase was caused to fall and subsequently break when the table moved.

Third, there are a number of likely inferences whose generation would be subject to individual differences. For instance, many readers would infer that the shards were located on the floor, but they could just as easily be thought of as being on the table.

Finally, there are many inferences that could add to the richness of a reader's understanding of the passage, which would not be related to maintaining coherence. For example, a reader could infer that Janet was upset or that Brutus would now be spending more time outdoors

Thurlow, R., & van den Broek, P.(1997). Automaticity and Inference Generation During Reading Comprehension, Reading & Writing Quarterly 13 (2) p.165-181.

So what does research say about these four types of inference?

These four classes of potential inferences can be broken into two distinct categories (Long, Seely, Oppy, & Golding, 1996; McKoon & Ratcliff, 1992; Vonk & Noordman, 1990). The first two, anaphoric and causal, are inferences that are required for maintaining coherence. The others, likely and enriching, are inferences that elaborate on the meaning of the text.

…the inferences described earlier as required for coherence are made automatically during the reading process, at least by more skilled readers. Research into the inferences we described as elaborative shows weaker and sometimes contradictory evidence as to whether they occur during reading.

In light of such studies, researchers are coming closer to agreeing on a "minimalist" view regarding which inferences are made automatically during reading (McKoon & Ratcliff, 1992). In this view, only those inferences that are required to maintain coherence are seen as being automatic.

Thurlow, R., & van den Broek, P.(1997). Automaticity and Inference Generation During Reading Comprehension, Reading & Writing Quarterly 13 (2) p.165-181.

And that leads to two specific instructional recommendations for teaching inference as a comprehension strategy:

First, it is apparent that skilled readers do have a goal of constructing a coherent memory representation. They make, at least, a minimum number of automatic inferences that help integrate new material with the previous text. If integration is difficult, they will take time to work out the difficulty with elaborations. A recommendation for reading instruction is to make sure developing readers are aware that the pieces of a text are meant to go together and that they, the readers, have a task to find out how the pieces fit.

Second, skilled readers have strategies and criteria for repairing coherence breaks. They re-read difficult passages; they look back in the text for more information; they even make up their own explanations that make sense. These are all teachable skills…

Thurlow, R., & van den Broek, P.(1997). Automaticity and Inference Generation During Reading Comprehension, Reading & Writing Quarterly 13 (2) p.165-181.

How kids learn to summarize text

Researchers Brown and Day examined the developmental sequence of six strategies students use to summarize text (my emphasis):

Two of the six rules involve deletion of unnecessary material. One should obviously delete material that is trivial, and even grade school children are quite adept at this if the form and content of the material is familiar (Brown, Day, & Jones, in press; Johnson, 1978). One should also delete material that, although it is important, is also redundant. Kintsch and van Dijk’s system also includes these two deletion rules. Two of the rules of summarization involve the substitution of a superordinate term or event for a list of items or actions. For example, if a text contains a list such as: cats, dogs, goldfish, gerbils, and parrots, one can substitute the term pets. This is Kintsch and van Dijk’s generalization rule. Similarly, one can substitute a superordinate action for a list of subcomponents of that action, i.e., John went to London, for: John left the house, John went to the train station, John bought a ticket, etc., etc. This is roughly comparable to Kintsch and van Dijk’s (1978) integration rule. The two remaining rules have to do with providing a summary of main constituent unit of text, the paragraph. The first rule is – select a topic sentence, if any, for this is the author’s summary of the paragraph. The final rule is – if there is no topic sentence, invent your own.

Brown, A.L., & Day, J.D. (1983). Macrorules for Summarizing Texts: The Development of Expertise. Center for the Study of Reading, Illinois University

What they found may help teachers identify the developmental stage a student is at:

“…a clear developmental pattern was found, with deletion rules emerging first followed by superordination [substitution] and then selection. Invention, the most difficult rule, was late developing. We believe that the five rules differ in their ease of application because they demand different degrees of text manipulation on the part of the learner, and perhaps because they depart to a greater or lesser extent from the already existing strategy favored by the younger participants. This has been called the copy-delete strategy (Brown, 1981; Brown, Day, & Jones, in press) because fifth and seventh grade and junior college students summarize texts primarily by deleting, or copying near verbatim the words actually present in the text

Brown, A.L., & Day, J.D. (1983). Macrorules for Summarizing Texts: The Development of Expertise. Center for the Study of Reading, Illinois University

Intriguingly, they suggest that an earlier, over-simple strategy may need to be actively unlearned in order to progress to the later strategies:

““…there is evidence that partially adequate strategies such as copy-delete are not just way-stations on the road to expert strategies; they may actually impede progress...We would like to argue that partially adequate strategies such as copy-delete and knowledge-telling are maintained by inexperienced writers because they do result in intermittent reinforcement and are recognizable attempts to get the job done. The process of development is not just one of acquiring increasingly more refined and sophisticated strategies; development involves the systematic consolidation of mature strategies, combined with the rejection of plausible but less efficient habits

Brown, A.L., & Day, J.D. (1983). Macrorules for Summarizing Texts: The Development of Expertise. Center for the Study of Reading, Illinois University

Building comprehension skills through distributed discussion

…the message of this study is not that any discussion that occurs during reading will likely be effective. Rather, the results of this study indicate that discussions of text that focus on meaning, are collaborative, and are under teacher guidance seem more effective if discussion occurs in a distributed manner.

Sandora, C., Beck, I., McKeown, M. (1999) A Comparison of Two Discussion Strategies on Students’ Comprehension and Interpretation of Complex Literature.Reading Psychology, 20:3, 177-212

By distributed the authors of this study mean stopping during the reading of a passage and discussing. The technique is often called "Questioning the Author". They propose two types of question that the teacher can use to instigate discussion:

Two kinds of probes, Initiating Queries and Follow-up Queries, have been conceptualized to assist teachers in handling discussion. Teachers use Initiating Queries, such as ‘‘What is the author trying to say?’’ and ‘‘What do you think the author means by that?’’ to begin a discussion. These Queries provide opportunities to make text content public so that discussion can ensue. Followup Queries, such as ‘‘Why do you think the author is telling us that?’’ and ‘‘How does that information connect to what the author discussed earlier?’’ were developed to assist the teacher in extending students’ responses and move discussion to deeper levels of engagement. Throughout discussion, students and teachers work collaboratively to construct meaning, as well as offer various interpretations of texts.

The study compared this type of discussion with the more typical format: teacher-led discussion after a text has been read.

The results of this study indicate that distributed discussion can facilitate students’ comprehension and interpretation of complex literature. Dealing with a complex narrative during the course of reading may be particularly effective for lower-achieving students. It can scaffold students’ comprehension processes by providing opportunities for students to re ect on events and ideas as they are encountered and to examine connections that accumulate. Often, students who have had less success in reading are not offered challenging pieces of literature in their language arts classes. Yet it is clear from this study that, at least when discussion occurs during the course of the reading, such students are well able to comprehend and interpret complex literature selections.