First, Do No Harm
To impart adequate verbal competence is the most important single goal of schooling in any nation. Verbal scores are reliable indexes to general competence, life chances, and civic participation. Good verbal scores diminish the notorious income gap. Decades of data show that the earnings gaps between racial and ethnic groups in the United States largely disappear when language competence in Standard English is factored in. But the verbal gaps remain large, and average scores of 17-year-olds have even declined. Considering the nugatory effect of state standards on verbal achievement, they must be judged a complete failure. Will the proposed new national standards work any better?
The authors of these “college ready” language standards, released in September, have done some things well. They say that students must command a “base of knowledge across a wide range of subject matter by engaging with works of quality and substance.” Also to be praised is their emphasis on the broad range of verbal activities necessary to language competence, seen in their main title to this section: “Reading, Writing, Speaking, and Listening.” This is an important conceptual advance over the less precise term “language arts,” which has long been restrictively associated with fiction and poetry. But literature is just one of the verbal arts, one among several domains that are critical to language proficiency. It does not warrant its current monopoly as the primary focus of instruction during the many hours that are devoted to language skill in the elementary grades.
Less admirable is the standards-drafters’ reliance on technical and formal descriptions of language proficiencies, thus repeating the error of current state standards of encouraging main-idea hunting and “inferencing.” There is no good scientific basis for believing that exercises in logical inference from texts or main-idea finding can significantly raise language abilities. Inference in language is not chiefly a formal skill. Untrained people are able to make very good inferences from texts when they already know something about the subject. But they cannot reliably draw correct inferences from texts about unfamiliar subjects.
In one well-known study, baseball-expert students judged “poor readers” based on standardized testing scored substantially higher than baseball-novice “good readers” on a reading test about baseball. (The same results have been obtained in Europe with texts on soccer.) These experimental results are not surprising when one grasps that most inference in language is not a formal, logical procedure, and cannot be much helped by practicing formal procedures. If the text says “John sacrificed and knocked in a run,” those with domain knowledge can construct the situation, and infer that there was probably a runner on third base—because they know baseball, not because they know logic or have been trained in main-idea finding.
At the very least, then, language standards need to say clearly and forcefully that standards in reading, writing, speaking, and listening are not intended to be explicitly taught as skills. Rather, even these preliminary standards need to stress that academic content—in literature, history, science, and the arts—must be taught coherently and cumulatively in order to impart the requisite language competencies. There is no other way to verbal competence. The formalistic approach has failed for many years and will continue to do so.
The great theorist of testing Samuel Messick said that school assessments should exhibit “consequential validity;” that is, they should encourage good teaching and learning. Writers of standards have the same obligation. As I have indicated, one reason the draft standards are not consequentially valid is that they encourage a scientifically inadequate view of language competence as a how-to skill. To this conceptual flaw must be added a related one—an equally formalistic (and equally evasive) conception of how the content of language courses should be determined.
The answer given in the draft “common-core standards” is that chosen texts must meet the criteria of “quality, range, and complexity,” with complexity being determined by the Qualitative Assessment of Text Difficulty devised by the late reading researcher Jeanne Chall, and by the quantitative Coh-Metrix method, which, according to its creators, “incorporates into its computer-basedanalysis more than 60 specific indices of syntax, semantics, readability, and cohesion to assess text complexity.”
The exemplary texts offered in the standards to illustrate the proper characteristics include such writers as Thomas Jefferson, Jane Austen, Walt Whitman, Toni Morrison, and Martin Luther King Jr.—all highly unlikely to have been chosen on computer-based grounds. Only an official commission would run the Declaration of Independence through a computer program to test whether it is suitable to be taught. Yet teachers are cautioned that the Declaration is syntactically demanding, with “little explicit cohesion between sentences—links supplied by words and phrases such as ‘for example,’ ‘moreover,’ or ‘in addition’—to help readers understand the relationship between the ideas being expressed.” I believe the most useful service I can offer to help improve these standards is to explain why a formalistic approach to language standards cannot be consequentially—or even scientifically—valid.
An extrinsic, formal approach to curricular choices cannot work, because key elements of complexity, like the key elements of inference, lie less in the formal characteristics of the text than in the contents of the reader’s or listener’s mind. The psychological difficulty of a text is determined less by its computer-measurable syntactical features than by the reader’s relevant prior knowledge. This was well illustrated during the heyday of mechanical readability formulas when a waggish scholar ran some of Gertrude Stein’s writings (such as “A rose is a rose is a rose”) through a quantitative program, and showed that her writings were suitable for 2nd graders—though few adults could understand them.
One text that is said to meet the criteria of “quality, range, and complexity” is Jane Austen’s Pride and Prejudice. The opening paragraph is quoted in the standards document to illustrate the difficulty of Austen’s syntax. Yet the first sentence of that paragraph is a brilliant example of syntactical simplicity. Its complexity doesn’t lie in its Coh-Metrix features, but in its computer-invisible irony:
“It is a truth universally acknowledged, that a single man in possession of a good fortune, must be in want of a wife.”
A computer program cannot score very well that which a text does not state. Like the implied runner on third base, irony subsists in what is unsaid. Among Austen’s unstated truths is the fact that a man’s want of a wife really has nothing to do with his having a good fortune, and that she doesn’t literally believe what she is saying, and that she assumes we readers don’t either, and, moreover, that we understand that she does not believe it. The result of comprehending the “main idea” is that the truth “universally acknowledged” is not a truth acknowledged by either the author or any ethically alert reader who has managed to understand the text. One could discuss further computer-invisible, syntax-indifferent complexities of that sentence, if space permitted.
The word “standards,” as used in recent years, is a vague word that has encouraged the public to believe that real curricular guides are being created, while enabling their writers to avoid making difficult but necessary curricular decisions that could guide the creators of classroom materials, teachers, and test-makers.
There are two ways in which makers of standards could overcome the political difficulties of performing their chief duty: giving useful guidance. One would be to offer one or more exemplary curriculum guides. For “college- and career-ready” verbal standards, it would mean grade-by-grade curricular guides from kindergarten through 12th grade. High school is too late. Language competence is a plant of slow growth. The seedling years are decisive—not the high school years. The exemplary guides would need to be K-12, specific, grade by grade, and to spread their tentacles into all the subject matters necessary to verbal competence in the modern world.
Another way of being actually useful would be to set forth in great detail the kind of criteria a local curriculum guide would have to fulfill to meet its pastoral obligations. Finland, like many other countries with high academic achievement, defines with great specificity what a school system must do to meet the goals set forth in its national standards. Every local school system must create a curriculum document that contains key specifics: a mission statement explaining the values behind its schooling, as well as its chief working methods; a description of how its specific courses are integrated with one another and how they reinforce specific cross-curricular themes; an account of how many lesson hours are to be devoted to each subject and topic; course descriptions stating the objectives for each course and its specific contents; and a specification of how the goals for each course and the general mission will be assessed.
Specific time allocations and specific grade-by-grade content requirements like those of Finland are necessary to cumulativeness in schooling. Within a grade, the content must be coherent, because language and knowledge growth is much more rapid when classrooms stay on a domain long enough to increase familiarity.
We Americans have had an allergy to tackling the content problem at any level—ignoring the fact that somebody (mainly textbook makers) must always be dictating content in the schools, even if it is trivial, fragmented, skills-based content. If the crafters of our standards don’t encourage or require content coherence and cumulativeness (just to name two necessary elements), they will have failed the most basic requirement of this task: First, do no harm. And they will have done little to improve the unacceptable stasis in American education.
Vol. 29, Issue 17, Page 29