Reading Scientific Text

There are many memory strategies that can be effective in improving your recall of text. However, recent research shows that it is simplistic to think that you can improve your remembering by applying any of these strategies to any text. Different strategies are effective with different types of text.

One basic classification of text structure would distinguish between narrative text and expository text. We are all familiar with narrative text (story-telling), and are skilled in using this type of structure. Perhaps for this reason, narrative text tends to be much easier for us to understand and remember. Most study texts, however, are expository texts.

Unfortunately, many students (perhaps most) tend to be blind to the more subtle distinctions between different types of expository structure, and tend to treat all expository text as a list of facts. Building an effective mental model of the text (and thus improving your understanding and recall) is easier, however, if you understand the type of structure you're dealing with, and what strategy is best suited to deal with it.

Identifying structure

Five common types of structure used in scientific texts are:

  • Generalization: the extension or clarification of main ideas through explanations or examples
  • Enumeration: listing of facts
  • Sequence: a connecting series of events or steps
  • Classification: grouping items into classes
  • Comparison / contrast: examining the relationships between two or more things

Let's look at these in a little more detail.

Generalization

In generalization, a paragraph always has a main idea. Other sentences in the paragraph either clarify the main idea by giving examples or illustrations, or extend the main idea by explaining it in more detail. Here's an example:

Enumeration

Enumeration passages may be a bulleted or numbered list, or a list of items in paragraph form, for example:

Sequence

A sequence describes a series of steps in a process. For example:

Classification

In classification, items are grouped into categories. For example:

Comparison / contrast

This type of text looks at relationships between items. In comparison, both similarities and differences are studied. In contrast, only the differences are noted. For example:

[examples taken from Cook & Mayer 1988]

A study [1] involving undergraduate students inexperienced in reading science texts (although skilled readers otherwise) found that even a small amount of training substantially improved the students' ability to classify the type of structure and use it appropriately.

Let's look briefly at the training procedures used:

Training for generalization

This involved the following steps:

  • identify the main idea
  • list and define the key words
  • restate the main idea in your own words
  • look for evidence to support the main idea
    • what kind of support is there for the main idea?
    • are there examples, illustrations?
    • do they extend or clarify the main idea?

Training for enumeration

This involved the following steps:

  • name the topic
  • identify the subtopics
  • organize and list the details within each subtopic, in your own words

Training for sequence

This involved the following steps:

  • identify the topic
  • name each step and outline the details within each
  • briefly discuss what's different from one step to another

[Only these three structures were covered in training]

Most effective text structures

Obviously, the type of structure is constrained by the material covered. We can, however, make the general statement that text that encourages the student to make connections is most helpful in terms of both understanding and memory.

In light of this, compare/contrast would seem to be the most helpful type of text. Another text structure that is clearly of a similar type has also been found to be particularly effective: refutational text. In a refutational text, a common misconception is directly addressed (and refuted). Obviously, this is only effective when there is a common misconception that stands in the way of the reader's understanding -- but it's surprising how often this is the case! Incompatible knowledge is at least as bad as a lack of knowledge in hindering the learning of new information, and it really does need to be directly addressed.

Refutational text is however, not usually enough on its own. While helpful, it is more effective if combined with other, supportive, strategies. One such strategy is elaborative interrogation, which involves (basically) the student asking herself why such a fact is true.

Unfortunately, however, text structures that encourage connection building are not the most common type of structure in scientific texts. Indeed, it has been argued that "the presentation of information in science textbooks is more likely to resemble that of a series of facts [and thus] presents an additional challenge that may thwart readers' efforts to organize text ideas relative to each other".

Most effective strategies

The fundamental rule (that memory and understanding are facilitated by any making of connections) also points to the strategies that are most effective.

As a general rule, strategies that involve elaborating the connections between concepts in a text are the most effective, but it is also true that the specifics of such strategies vary according to the text structure (and other variables, such as the level of difficulty).

Let's look at how such a linking strategy might be expressed in the context of our five structures.

Generalization

Restatement in your own words -- paraphrasing -- is a useful strategy not simply because it requires you to actively engage with the material, but also because it encourages you to connect the information to be learned with the information you already have in your head. We can, however, take this further in the last stage, when we look for the evidence supporting the main idea, if we don't simply restrict ourselves to the material before us, but actively search our minds for our own supporting evidence.

Enumeration

This text structure is probably the hardest to engage with. You may be able to find a connective thread running through the listed items, or be able to group the listed items in some manner, but this structure is the one most likely to require mnemonic assistance (see verbal mnemonics and list-learning mnemonics).

Sequence

With this text structure, items are listed, but there is a connecting thread — a very powerful one. Causal connections are ones we are particularly disposed to pay attention to and remember; they are the backbone of narrative text. So, sequence has a strong factor going for it.

Illustrations particularly lend themselves to this type of structure, and research has shown that memory and comprehension is greatly helped when pictures portraying a series of steps, in a cause-and-effect chain, are closely integrated with explanatory text. The closeness is vital — a study that used computerized instruction found dramatic improvement in memory when the narration was synchronous with the animation, for example, but there was no improvement when the narration was presented either before or after the text. If you are presented with an illustration that is provided with companion text, but is not closely integrated with it, you will probably find it helpful to integrate it with the text yourself.

Classification

Classification is frequently as simple as grouping items. However, while this is in itself a useful strategy that helps memory, it will be more effective if the connections between and within groups are strong and clear. Connections within groups generally emphasize similarities, while connections between groups emphasize both similarities (between closely connected groups) and differences. Ordering groups in a hierarchical system is probably the type of arrangement most familiar to students, but don't restrict yourself to it. Remember, the important thing is that the arrangement has meaning for you, and that the connections emphasize the similarities and differences.

Compare / contrast

This type of structure lends itself, of course, to making connections. Your main strategy is probably therefore to simply organize the material in such a way as to make those connections clear and explicit.

References: 
  1. Cook, L.K. & Mayer, R.E. 1988. Teaching readers about the structure of scientific text. Journal of Educational Psychology, 80, 448-54.
  2. Castaneda, S., Lopez, M. & Romero, M. 1987. The role of five induced learning strategies in scientific text comprehension. The Journal of Experimental Education, 55(3), 125–131.
  3. Diakidoy, I.N., Kendeou, P. & Ioannides, C. 2002. Reading about energy: The effects of text structure in science learning and conceptual change. http://www.edmeasurement.net/aera/papers/KENDEOU.PDF

Working memory, expertise & retrieval structures

In a 1987 experiment (1), readers were presented with a text that included one or other of these sentences:

After doing a few warm-up exercises, John put on his sweatshirt and began jogging.

or

After doing a few warm-up exercises, John took off his sweatshirt and began jogging.

Both texts went on to say: John jogged halfway around the lake.

After reading the text, readers were asked if the word sweatshirt had appeared in the story. Now here is the fascinating and highly significant result: those who read that John had put on a sweatshirt responded “yes” more quickly than those who had read that he had taken off his sweatshirt.

Why is this so significant? Because it tells us something important about the reading process, at least in the minds of skilled readers. They construct mental models. If it was just a matter of the mechanical lower-order processing of letters and words, why would there be a difference in responses? Neither text was odd — John could as well have put on a sweatshirt before going out for a jog as taken it off — so there shouldn’t be a surprise effect. So what is it? Why is the word sweatshirt not as tightly / strongly linked in the second case as it is in the first? If they were purely textbase links (links generated by the textbase itself), the links should be equivalent. The difference in responses implies that the readers are making links with something outside the textbase, with a mental model.

Mental models, or as they are sometimes called in this context, situation models, are sometimes represented as lists of propositions, but in most cases it seems likely that they are actually analogue in nature. Thus the real world should be better represented by the situation model than by the text. Moreover, a spatial situation model will be similar in many ways to an image, with all the advantages that that entails.

All of this has relevance to two very important concepts: working memory and expertise.

Now, I’m always talking about working memory. This time I want to discuss not so much the limited attentional capacity that is what we chiefly mean by working memory, but another, more theoretical concept: the idea of long-term working memory.

Think about reading. To make sense of the text you need to remember what’s gone before — this is why working memory is so important for the reading process. But we know how limited working memory is; it can only hold a very small amount — is it really possible to hold all the information we need to make sense of what we’re reading? Shouldn’t there be constant delays as we access needed information from long-term memory? But there aren’t.

It’s suggested that the answer lies in the use of long-term working memory, a retrieval structure that keeps a network of linked propositions readily available.

Think about when you are studying / reading a difficult text in a subject you know well. Compare this to studying a difficult text in a subject you don’t know well. In the latter case, you may have to painfully backtrack, checking earlier statements, trying to remember what was said before, trying to relate what you are reading to things you already know. In the former case, you seem to have a vastly expanded amount of readily accessible relevant information, from the text itself and from your long-term memory.

The connection between long-term working memory and expertise is obvious. And expertise has already been conceptualised in terms of retrieval structures (see for example my article on expertise). In other words, you can increase your working memory in a particular domain by developing expertise, and the shortest route to developing expertise is to concentrate on building effective retrieval structures.

One of the areas where this is particularly crucial is that of reading scientific texts. Now we all know that scientific texts are much harder to process than, for example, stories. And there are several reasons for that. One is the issue of language: any science has its own technical vocabulary and you won't get far without knowing it. But another reason, far less obvious to the untutored, concerns the differences in structure — what may be termed differences of genre.

Now it might seem self-evident that stories are far simpler than science, than any non-fiction texts, and indeed a major distinction is usually made between narrative texts and expository texts, but it’s rather like the issue of faces and other objects. Are we specially good at faces because we're 'designed' to be (i.e., we have special 'expert' modules for processing faces)? Or is it simply that we have an awful lot of practice at it, because we are programmed to focus on human faces almost as soon as we are born?

In the same way, we are programmed for stories: right from infancy, we are told stories, we pay attention to stories, we enjoy stories. Stories have a particular structure (and within the broad structure, a set of sub-structures), and we have a lot of practice in that structure. Expository texts, on the other hand, don't get nearly the same level of practice, to the extent that many college students do not know how to handle them — and more importantly, don't even realize that that is what they're missing: a retrieval structure for the type of text they're studying.

References: 

Glenberg, A.M., Meyer, M. & Lindem, K. 1987. Mental models contribute to foregrounding during text comprehension. Journal of Memory and Language, 26, 69-83.

Understanding scientific text

In the last part I talked about retrieval structures and their role in understanding what you’re reading. As promised, this month I’m going to focus on understanding scientific text in particular, and how it differs from narrative text.

First of all, a reminder about situation models. A situation, or mental, model is a retrieval structure you construct from a text, integrating the information in the text with your existing knowledge. Your understanding of a text depends on its coherence; it’s generally agreed that for a text to be coherent it must be possible for a single situation model to be constructed from it (which is not to say a text that is coherent is necessarily coherent for you —that will depend on whether or not you can construct a single mental model from it).

There are important differences in the situation models constructed for narrative and expository text. A situation model for a narrative is likely to refer to the characters in it and their emotional states, the setting, the action and sequence of events. A situation model for a scientific text, on the other hand, is likely to concentrate on the components of a system and their relationships, the events and processes that occur during the working of the system, and the uses of the system.

Moreover, scientific discourse is rooted in an understanding of cause-and-effect that differs from our everyday understanding. Our everyday understanding, which is reflected in narrative text, sees cause-and-effect in terms of goal structures. This is indeed the root of our superstitious behavior — we (not necessarily consciously) attribute purposefulness to almost everything! But this approach is something we have to learn not to apply to scientific problems (and it requires a lot of learning!).

This is worth emphasizing: science texts assume a different way of explaining events from the way we are accustomed to use — a way that must be learned.

In general, then, narrative text (and ‘ordinary’ thinking) is associated with goal structures, and scientific text with logical structures. However, it’s not quite as clear-cut a distinction as all that. While the physical sciences certainly focus on logical structure, both the biological sciences and technology often use goal structures to frame their discussions. Nevertheless, as a generalization we may say that logical thinking informs experts in these areas, while goal structures are what novices focus on.

This is consistent with another intriguing finding. In a comparison of two types of text —ones discussing human technology, and ones discussing forces of nature — it was found that technological texts were more easily processed and remembered. Indications were that different situation models were constructed — a goal-oriented representation for the technological text, and a causal chain representation for the force of nature text. The evidence also suggested that people found it much easier to make inferences (whether about agents or objects) when human agents were involved. Having objects as the grammatical subject was clearly more difficult to process.

Construction of the situation model is thus not solely determined by comprehension difficulty (which was the same for both types of text), but is also affected by genre and surface characteristics of the text.

There are several reasons why goal-oriented, human-focused discourse might be more easily processed (understood; remembered) than texts describing inanimate objects linked in a cause-effect chain, and they come down to the degree of similarity to narrative. As a rule of thumb, we may say that to the degree that scientific text resembles a story, the more easily it will be processed.

Whether that is solely a function of familiarity, or reflects something deeper, is still a matter of debate.

Inference making is crucial to comprehension and the construction of a situation, because a text never explains every single word and detail, every logical or causal connection. In the same way that narrative and expository text have different situation models, they also involve a different pattern of inference making. For example, narratives involve a lot of predictive inferences; expository texts typically involve a lot of backward inferences. The number of inferences required may also vary.

One study found that readers made nine times as many inferences in stories as they did in expository texts. This may be because there are more inferences required in narratives — narratives involve the richly complex world of human beings, as opposed to some rigidly specified aspect of it, described according to a strict protocol. But it may also reflect the fact that readers don’t make all (or indeed, anywhere near) the inferences needed in expository text. And indeed, the evidence indicates that students are poor at noticing coherence gaps (which require inferences).

In particular, readers frequently don’t notice that something they’re reading is inconsistent with something they already believe. Moreover, because of the limitations of working memory, only some of the text can be evaluated for coherence at one time (clearly, the greater the expertise in the topic, the more information that can be evaluated at one time — see the previous newsletter’s discussion of long-term working memory). Less skilled (and younger) readers in particular have trouble noticing inconsistencies within the text if they’re not very close to each other.

Let’s return for a moment to this idea of coherence gaps. Such gaps, it’s been theorized, stimulate readers to seek out the necessary connections and inferences. But clearly there’s a particular level that is effective for readers, if they often miss them. This relates to a counter-intuitive finding — that it’s not necessarily always good for the reader if the text is highly coherent. It appears that when the student has high knowledge, and when the task involves deep comprehension, then low coherence is actually better. It seems likely that knowledgeable students reading a highly coherent text will have an “illusion of competence” that keeps them from processing the text properly. This implies that there will be an optimal level of coherence gaps in a text, and this will vary depending on the skills and knowledge base of the reader.

Moreover, the comprehension strategy generally used with simple narratives focuses on referential and causal coherence, but lengthy scientific texts are likely to demand more elaborate strategies. Such strategies are often a problem for novices because they require more knowledge than can be contained in their working memory. Making notes (perhaps in the form of a concept map) while reading can help with this.

Next month I’ll continue this discussion, with more about the difficulties novices have with scientific texts and what they or their teachers can do about it, and the problems with introductory textbooks. In the meantime, the take-home message from this is:

Understanding scientific text is a skill that must be learned;

Scientific text is easier to understand the more closely it resembles narrative text, with a focus on goals and human agents;

How well the text is understood depends on the amount and extent of the coherence gaps in the text relative to the skills and domain knowledge of the reader.

References: 

Otero, J., León, J.A. & Graesser, A.C. (eds). 2002. The psychology of science text comprehension.

Novices' problems with scientific text

This is the last part in my series on understanding scientific text. In this part, as promised, I am going to talk about the difficulties novices have with scientific texts; what they or their teachers can do about it; and the problems with introductory textbooks.

The big problem for novices is of course that their lack of knowledge doesn’t allow them to make the inferences they need to repair the coherence gaps typically found in such texts. This obviously makes it difficult to construct an adequate situation model. Remember, too, that to achieve integration of two bits of information, you need to have both bits active in working memory at the same time. This, clearly, is more difficult for those for whom all the information is unfamiliar (remember what I said about long-term working memory last month).

But it’s not only a matter a matter of having knowledge of the topic itself. A good reader can compensate for their lack of relevant topic knowledge using their knowledge about the structure of the text genre. For this, the reader needs not only to have knowledge of the various kinds of expository structures, but also of the cues in the text that indicate what type of structure it is. (see my article on Reading scientific text for more on this).

One of the most effective ways of bringing different bits of information together is through the asking of appropriate questions. Searching a text in order to answer questions, for example, is an effective means of improving learning. Answering questions is also an effective means of improving comprehension monitoring (remember that one of the big problems with reading scientific texts is that students tend to be poor at judging how well they have understood what was said).

One of the reasons why children typically have pronounced deficits in their comprehension monitoring skills when dealing with expository texts, is that they have little awareness that expository texts require different explanations than narrative texts. However, these are trainable skills. One study, for example, found that children aged 10-12 could be successfully taught to use “memory questions” and “thinking questions” while studying expository texts.

Moreover, the 1994 study found that when the students were trained to ask questions intended to access prior knowledge/experience and promote connections between the lesson and that knowledge, as well as questions designed to promote connections among the ideas in the lesson, their learning and understanding was better than if they were trained only in questions aimed at promoting connections between the lesson ideas only (or if they weren’t trained in asking questions at all!). In other words, making explicit connections to existing knowledge is really important! You shouldn’t just be content to consider a topic in isolation; it needs to be fitted into your existing framework.

College students, too, demonstrate limited comprehension monitoring, with little of their self-questioning going deeply into the material. So it may be helpful to note Baker’s 7 comprehension aspects that require monitoring:

  1. Your understanding of the individual words
  2. Your understanding of the syntax of groups of words
  3. External consistency — how well the information in the text agrees with the knowledge you already have
  4. Internal consistency — how well the information in the text agrees with the other information in the text
  5. Propositional cohesiveness — making the connections between adjacent propositions
  6. Structural cohesiveness —integrating all the propositions pertaining to the main theme
  7. Information completeness — how clear and complete the information in the text is

Think of this as a checklist, for analyzing your (or your students’) understanding of the text.

But questions are not always the answer. The problem for undergraduates is that although introductory texts are presumably designed for novices, the students often have to deal not only with unfamiliar content, but also an approach that is unfamiliar. Such a situation may not be the best context for effective familiar strategies such as self-explanation.

It may be that self-explanation is best for texts that in the middle-range for the reader — neither having too little relevant knowledge, or too much.

Introductory texts also are likely to provide only partial explanations of concepts, a problem made worse by the fact that the novice student is unlikely to realize the extent of the incompleteness. Introductory texts also suffer from diffuse goals, an uneasy mix of establishing a basic grounding for more advanced study, and providing the material necessary to pass immediate exams.

A study of scientific text processing by university students in a natural situation found that the students didn’t show any deep processing, but rather two kinds of shallow processing, produced by either using their (limited knowledge of) expository structures, or by representing the information in the text more precisely.

So should beginning students be told to study texts more deeply? The researchers of this study didn’t think so. Because introductory texts suffer from these problems I’ve mentioned, in particular that of incomplete explanations, they don’t lend themselves to deep processing. The researchers suggest that what introductory texts are good for is in providing the extensive practice needed for building up knowledge of expository structures (and hopefully some necessary background knowledge of the topic! Especially technical language).

To that end, they suggest students should be advised to perform a variety of activities on the text that will help them develop their awareness of the balance between schema and textbase, with the aim of developing a large repertory of general and domain-specific schemata. Such activities / strategies include taking notes, rereading, using advance organizers, and generating study questions. This will all help with their later construction of good mental models, which are so crucial for proper understanding.

References: 
  • Baker, L. 1985. Differences in the standards used by college students to evaluate their comprehension of expository prose. Reading Research Quarterly, 20 (3), 297-313.
  • Elshout-Mohr, M. & van Daalen-Kapteijns, M. 2002. Situated regulation of scientific text processing. In Otero, J., León, J.A. & Graesser, A.C. (eds). The psychology of science text comprehension. Pp 223-252. Mahwah, NJ: LEA.
  • King, A. 1994. Guiding Knowledge Construction in the Classroom: Effects of Teaching Children How to Question and How to Explain. American Educational Research Journal, 31 (2), 338-368.