Comprehension Instruction: What Works

Reading is often thought of as a hierarchy of skills, from processing of individual letters and their associated sounds to word recognition to text-processing competencies. Skilled comprehension requires fluid articulation of all these processes, beginning with the sounding out and recognition of individual words to the understanding of sentences in paragraphs as part of much longer texts. There is instruction at all of these levels that can be carried out so as to increase student understanding of what is read.

Based on research, a strong case can be made for doing the following in order to improve reading comprehension in students:

• Teach decoding skills
• Teach vocabulary
• Word knowledge: Encourage students to build world knowledge through reading and to relate what they know to what they read (e.g., by asking “Why?” questions about factual knowledge in text).
• Active comprehension strategies: Teach students to use a repertoire of active comprehension strategies, including prediction, analyzing stories with respect to story grammar elements, question asking, image construction, and summarizing.
• Monitoring: Encourage students to monitor their comprehension, noting explicitly whether decoded words make sense and whether the text itself makes sense. When problems are detected, students should know that they need to reprocess (e.g., by attempting to sound out problematic words again or rereading).

Encourage students to monitor their comprehension, noting explicitly whether decoded words make sense and whether the text itself makes sense. When problems are detected, students should know that they need to reprocess (e.g., by attempting to sound out problematic words again or rereading).

Such instruction must be long term, for there is much to teach and much for young readers to practice. Even so, there is little doubt that instruction that develops these interrelated skills should improve comprehension.

Decoding

Perhaps it is a truism, but students cannot understand texts if they cannot read the words. Before they can read the words, they have to be aware of the letters and the sounds represented by letters so that sounding out and blending of sounds can occur to pronounce words (see, e.g., Nicholson, 1991). Once pronounced, the good reader notices whether the word as recognized makes sense in the sentence and the text context being read and, if it does not, takes another look at the word to check if it might have been misread (e.g., Gough, 1983, 1984). Of course, reading educators have paid enormous attention to the development of children’s word-recognition skills because they recognize that such skills are critical to the development of skilled comprehenders.

As part of such work, LaBerge and Samuels (1974) made a fundamental discovery. Being able to sound out a word does not guarantee that the word will be understood as the child reads. When children are first learning to sound out words, it requires real mental effort. The more effort required, the less consciousness left over for other cognitive operations, including comprehension of the words being sounded out. Thus, LaBerge and Samuels’ analyses made clear that it was critical for children to develop fluency in word recognition. Fluent (i.e., automatic) word recognition consumes little cognitive capacity, freeing up the child’s cognitive capacity for understanding what is read. Anyone who has ever taught elementary children and witnessed round-robin reading can recall students who could sound out a story with great effort but at the end had no idea of what had been read.

Tan and Nicholson (1997) carried out a study that emphasized the importance of word-recognition instruction to the point of fluency. In their study, struggling primary-level readers were taught 10 new words, with instruction either emphasizing word recognition to the point of fluency (they practiced reading the individual words until they could recognize them automatically) or understanding of the words (instruction involving mostly student-teacher discussions about word meanings). Following the instruction, the students read a passage containing the words and answered comprehension questions about it. The students who had learned to recognize the words to the point of automaticity answered more comprehension questions than did students who experienced instruction emphasizing individual word meanings. Consistent with other analyses (e.g., Breznitz, 1997a, 1997b), Tan and Nicholson’s outcome made obvious that development of fluent word-recognition skills can make an important difference in students’ understanding of what they read.

Thus, a first recommendation to educators who want to improve students’ comprehension skills is to teach them to decode well. Explicit instruction in sounding out words, which has been so well validated as helping many children to recognize words more certainly (e.g., Snow, Burns, & Griffin, 1998, online document), is a start in developing good comprehenders – but it is just a start. Word-recognition skills must be developed to the point of fluency if comprehension benefits are to be maximized.

Vocabulary

It is well established that good comprehenders tend to have good vocabularies (Anderson & Freebody, 1991; Nagy, Anderson, & Herman, 1987). This correlation, however, does not mean that teaching vocabulary will increase readers’ comprehension, for that is a causal conclusion. As it turns out, however, when reading educators conducted experiments in which vocabulary was either taught to students or not, comprehension improved as a function of vocabulary instruction. Perhaps the most widely cited experiment of this type was carried out by Isabel Beck and her associates, who taught Grade 4 children a corpus of 104 words over a 5-month period (Beck, Perfetti, & McKeown, 1982). The children who received instruction outperformed noninstructed children on subsequent comprehension tests. When all of the work of Beck’s group and others is considered (see, e.g., Beck & McKeown, 1991; Durso & Coggins, 1991), a good case can be made that when students are taught vocabulary in a thorough fashion, their comprehension of what they read improves.

One counterargument to this advice to teach vocabulary is that children learn vocabulary incidentally – that is, they learn the meanings of many words by experiencing those words in the actual world and in text worlds, without explicit instruction (Stanovich, 1986; Sternberg, 1987). Even so, such incidental learning is filled with potential pitfalls, for the meanings learned range from richly contextualized and more than adequate to incomplete to wrong (Miller & Gildea, 1987). Just the other morning, I sat in a reading class as a teacher asked students to guess the meanings of new words encountered in a story, based on text and picture clues. Many of the definitions offered by the children were way off. Anyone who has ever taught young children knows that they benefit from explicit teaching of vocabulary.

That children do develop knowledge of vocabulary through incidental contact with new words they read is one of the many reasons to encourage students to read extensively. Whenever researchers have looked, they have found vocabulary increases as a function of children’s reading of text rich in new words (e.g., Dickinson & Smith, 1994; Elley, 1989; Morrow, Pressley, Smith, & Smith, 1997; Pelligrini, Galda, Perlmutter, & Jones, 1994; Robbins & Ehri, 1994; Rosenhouse, Feitelson, Kita, & Goldstein, 1997).

World knowledge

Reading comprehension can be affected by world knowledge, with many demonstrations that readers who possess rich prior knowledge about the topic of a reading often understand the reading better than classmates with low prior knowledge (Anderson & Pearson, 1984). That said, readers do not always relate their world knowledge to the content of a text, even when they possess knowledge relevant to the information it presents. Often, they do not make inferences based on prior knowledge unless the inferences are absolutely demanded to make sense of the text (McKoon & Ratcliff, 1992).

The received wisdom in recent decades, largely based on the work of Richard C. Anderson, P. David Pearson, and their colleagues at the Center for the Study of Reading at the University of Illinois in the 1970s, 1980s, and into the early 1990s, was that reading comprehension can be enhanced by developing reader’s prior knowledge. One way to accomplish this is to encourage extensive reading of high-quality, information-rich texts by young readers (e.g., Stanovich & Cunningham, 1993).

Typically, however, when readers process text containing new factual information, they do not automatically relate that information to their prior knowledge, even if they have a wealth of knowledge that could be related. In many cases, more is needed for prior knowledge to be beneficial in reading comprehension. A large number of experiments conducted in the late 1980s and early 1990s demonstrated the power of “Why?” questions, or “elaborative interrogation,” to encourage readers to orient to their prior knowledge as they read (Pressley, Wood, Woloshyn, Martin, King, & Menke, 1992). In these studies, readers were encouraged to ask themselves why the facts being presented in text made sense. This encouragement consistently produced a huge effect on memory of the texts, with the most compelling explanation emerging from analytical experiments being that the interrogation oriented readers to prior knowledge that could explain the facts being encountered (see especially Martin & Pressley, 1991). The lesson that emerged from these studies is that readers should be encouraged to relate what they know to information-rich texts they are reading, with a potent mechanism for doing this being elaborative interrogation.

Active comprehension strategies

Good readers are extremely active as they read, as is apparent whenever excellent adult readers are asked to think aloud as they go through text (Pressley & Afflerbach, 1995). Good readers are aware of why they are reading a text, gain an overview of the text before reading, make predictions about the upcoming text, read selectively based on their overview, associate ideas in text to what they already know, note whether their predictions and expectations about text content are being met, revise their prior knowledge when compelling new ideas conflicting with prior knowledge are encountered, figure out the meanings of unfamiliar vocabulary based on context clues, underline and reread and make notes and paraphrase to remember important points, interpret the text, evaluate its quality, review important points as they conclude reading, and think about how ideas encountered in the text might be used in the future. Young and less skilled readers, in contrast, exhibit a lack of such activity (e.g., Cordón & Day, 1996).

Reading researchers have developed approaches to stimulating active reading by teaching readers to use comprehension strategies. Of the many possible strategies, the following often produce improved memory and comprehension of text in children: generating questions about ideas in text while reading; constructing mental images representing ideas in text; summarizing; and analyzing stories read into story grammar components of setting, characters, problems encountered by characters, attempts at solution, successful solution, and ending (Pearson & Dole, 1987; Pearson & Fielding, 1991; Pressley, Johnson, Symons, McGoldrick, & Kurita, 1989).

Of course, excellent readers do not use such strategies one at a time, nor do they use them simply when under strong instructional control – which was the situation in virtually all investigations of individual strategies. Hence, researchers moved on to teaching students to use the individual strategies together, articulating them in a self-regulated fashion (i.e., using them on their own, rather than only on cue from the teacher). In general, such packages proved teachable, beginning with reciprocal teaching, the first such intervention (Palincsar & Brown, 1984), and continuing through more flexible approaches that began with extensive teacher explanation and modeling of strategies, followed by teacher-scaffolded use of the strategies, and culminating in student self-regulated use of the strategies during regular reading (e.g., Anderson, 1992; Brown, Pressley, Van Meter, & Schuder, 1996; Duffy et al., 1987). The more recent, more flexible form of this instruction came to be known as transactional strategies instruction (Pressley et al., 1992), with the body of research on this approach recently cited by the National Reading Panel (2000) as exemplary work in comprehension instruction. When such instruction has been successful, it has always been long term, occurring over a semester or school year at minimum, with consistent and striking benefits.

The case is very strong that teaching elementary, middle school, and high school students to use a repertoire of comprehension strategies increases their comprehension of text. Teachers should model and explain comprehension strategies, have their students practice using such strategies with teacher support, and let students know they are expected to continue using the strategies when reading on their own. Such teaching should occur across every school day, for as long as required to get all readers using the strategies independently – which means including it in reading instruction for years.

Monitoring

Good readers know when they need to exert more effort to make sense of a text. For example, they know when to expend more decoding effort – they are aware when they have sounded out a word but that word does not really make sense in the context (Isakson & Miller, 1976). When good readers have that feeling, they try rereading the word in question. It makes sense to teach young readers to monitor their reading of words in this way (Baker & Brown, 1984). Contemporary approaches to word-recognition instruction also include a monitoring approach, with readers taught to pay attention to whether the decoding makes sense and to try decoding again when the word as decoded is not in synchrony with other ideas in the text and pictures (e.g., Iversen & Tunmer, 1993).

Good readers are also aware of the occasions when they are confused, when text does not make sense (Baker & Brown, 1984). A key component in transactional strategies instruction is monitoring. Even the first such package, reciprocal teaching (Palincsar & Brown, 1984), included the clarification strategy: When readers did not understand a text, they were taught to seek clarification, often through rereading. To improve children’s reading and comprehension, it makes very good sense to teach them to monitor as they read, to ask themselves consistently, “Is what I am reading making sense?” Children also need to be taught that they can do something about it when text seems not to make sense: At a minimum, they can try sounding out a puzzling word again or rereading the part of a text that seems confusing.