The Point

The responses I received to my vocabulary comment were not contradicted by the research I presented . In fact, they could be supported by it.

And there are other reasons to call out vocabulary in a text. Here's Robert with one of them:

If my calculus students encountered a problem involving relative maximum values of a function, the first thing they need is to be fluent with the meaning of the term "relative maximum." If they're not, then they need to go look it up--and the cues show them where to look (by flagging the point at which the term is first introduced).

I agree that highlighting important vocabulary words helps signal their importance and can help with navigating a search for their meaning. But, to me, the idea should be to also help students remember the meaning of the words. In truth, publishers don't highlight vocabulary words because doing so facilitates recall of those words or because it signals their importance or because it helps with textbook navigation. They highlight vocabulary words because it signals to customers that the publisher cares about vocabulary. The point is that if we truly cared about learning, we could do more with text to facilitate learning.

Consider the following series of experiments conducted by Bruce Britton (whose work we have looked at before) and Robert Sorrells out of the University of Georgia. They began with questions that are rarely, if ever, asked in educational publishing:

Our development of the method began when we encountered Colomb and Williams' (1987) claim that all expository texts have a point, which is a central tenet around which the text is organized. From this idea, our three hypotheses emerged. First, do expert readers generally agree about the point of a text, whereas less skilled readers differ in their judgments of the point? Second, do readers who have the point of a text signaled to them recall more of that text than readers who do not have the point signaled? Finally, do readers who have the point signaled to them construct a different structure of the text, incorporating more of the important information than those who do not? If these hypotheses are supported, they suggest that identifying the point of a text and informing readers that it is the point would improve the learnability of that text.

Experiment 1

To answer the first question, the researchers assembled five faculty with Ph.D.s in English as the "expert readers" and 12 introductory psychology students as the "less skilled readers." Each individual was given three brief expository texts, without titles, and asked to simply underline the point of each after a careful read. The texts were provided by Colomb, who identified the point of each to the researchers (point units) along with the "themes related to the point." The "themes" (theme units) were ideas that were not as informative as the point, but were more important than other ideas (other idea units) in the texts. Each text was a different length, and "the point" in each text occurred in a different location.

The researchers found that the "expert readers" identified 86.4% of the points, whereas the "less skilled readers" succeeded only in identifying 49.5% of the points, a significant difference (p < 0.05). Less skilled readers also showed greater variance in their individual results.

The importance of this result is that it points to the possibility of experts signaling the point for less skilled readers. The next two experiments focus on whether or not such signaling can improve recall and/or comprehension. In these experiments, Britton and Sorrells set out to test this general hypothesis:

If the point of a text telegraphs the information in that text, as Colomb and Williams suggested, then cuing the point should affect the recall of other information.

Experiment 2

This experiment was conducted to answer the second and third of the researchers' specific questions:

Do readers who have the point of a text signaled to them recall more of that text than readers who do not have the point signaled? [And], do readers who have the point signaled to them construct a different structure of the text, incorporating more of the important information than those who do not?

Each of twenty-nine introductory psychology students was given the three texts used in Experiment 1. (Recall that each text was broken down by the researchers into "point units," "theme units," and "other idea units.") Each participant was asked to read the text and write down what he or she remembered (free recall). For participants in the experimental group, the point of each text was underlined, and written instructions identified the underlined sentence as the point of the text. Participants in the control group received the texts without modification of any kind. Reading time and recall time were measured.

[Note: There is no indication in the write-up as to whether participants were asked to read all three texts before recall or whether they were asked to recall after each text. However, in this experiment, the mean reading times ranged from 7.77 min to 9.08 min and the mean recall times ranged from 10.53 min to 12.04 min. The longest text in the experiment was 559 words. It seems unlikely that it would take about 8 to 9 minutes for college students to read such a brief passage and then spend 11 to 12 minutes writing down everything they could remember about it. But, hey, I've heard crazier things.]

Not only was overall recall significantly improved in the experimental condition--in which the point was underlined--but recall in every subgroup—point units, theme units, and idea units—was significantly improved:

First, the experimental group recalled significantly more of the points than the control group . . . Second, a subset of data was analyzed for differences between the groups for recall of the important themes (as identified by Colomb). The experimental group also recalled significantly more of these theme-related units than the control group . . . Also, the differences between the groups in recall for the unsignaled content (the themes and the other units) were assessed. The experimental group recalled more of these nonpoint units than the control group. [(p < 0.05)]

Importantly, no significant differences were found between the experimental and control groups for either reading time or recall time, which indicates that the underlining and extra instructions for the experimental group did not cause that group to spend more time processing the text or to take the task more seriously.

The results from the first two experiments suggest that not only is identifying the point possible, it is also beneficial for recall.

Experiment 3

This experiment replicated Experiment 2 and added another condition in which a random sentence was underlined and designated as the point. If the mere underlining of a sentence was responsible for the effects found in Experiment 2, then the random underlining condition would exhibit recall patterns similar to those found for the experimental group in Experiment 2. However, if the pattern of recall for the random group was similar to that for the control group of Experiment 2, then evidence would be provided that it is the semantic importance or informative content of the point that, when signaled, facilitated recall.

This experiment, conducted on forty-one introductory psychology students, found the same results as those in Experiment 2, while also finding no difference between the control group and the random group, indicating that signaling the actual point—not just signaling a random sentence and calling it the point—was responsible for the improved recall. Furthermore,

It is also interesting to note that in Experiment 3, the control group recalled more of the actual points, as well as more of the related themes than the random group (although these differences were not significant). If this particular finding is replicated, then an argument could be made that signaling random content can be detrimental to recall.

Of course, these experiments were conducted on college students, and, while the texts were expository, none of them were about mathematics. Still, the results suggest that, should we take seriously the idea of vocabulary instruction in mathematics (among other things), we should make new vocabulary--the words and their definitions--the explicit "point" of instruction and signal this point to students. These results also suggest that authors should rethink how they present content in textbooks. Rather than finding ever more clever ways to distract and "engage" students, they could spend some thought on what the "point" of the instruction is and build from there.


Reference:Learning from Text across Conceptual Domains. Contributors: Cynthia R. Hynd - editor. Publisher: Lawrence Erlbaum Associates. Mahwah, NJ (1998)