KISS Homepage
Dr. V's Main Menu
KISS Curriculum
Return to the 

Why the Anti-Grammarians are Wrong:
The Problems with Previous Research on Grammar

     This essay is the original draft of Chapter 2 ("Why the Anti-Grammarians are Wrong: The Problems with Previous Research") of  KISS Grammar (You'll Love It), a TRIP book written under contract for NCTE. After I wrote this draft, Peter Feely (my NCTE editor) and I agreed that the TRIP book should focus on the research that supports teaching grammar and that this draft should be put here on the web site for those who still believe that research "proves" that teaching grammar is harmful. In placing it here, I have made some revisions.

      "Judas went out and hanged himself. Go and do ye likewise." The statement was rather shocking. It came from a middle-aged priest who, we had thought, at least until that time, liked and respected us. It was ninth grade, religion class in a Catholic school. Thin and tall, Father McElheny towered above us as he stood in the middle of the room, calmly surveying the effect of his words. He certainly knew how to get our attention. It turned out, however, that he still loved us. The lesson was about the study of the Bible and about how some people distort that study by taking quotations out of context. His statement had been merely (?) an example. The lesson, however, is also relevant to the research about the effectiveness of teaching grammar. Too many people (within the profession) distort the research by citing conclusions without explaining the context, perhaps even without actually reading the research..
     A year or so ago, a parent who had found my web site sent me a rather humorous, but also very disturbing question. He was upset because his child's teacher was teaching the students to begin sentences with prepositional phrases. He had, he wrote, been taught in school that one should never begin a sentence with a prepositional phrase. In all probability, this parent confused "prepositional phrase" with "but," in that totally unfounded rule about not beginning sentences with them. But whatever the case, I found the situation absurd, and thus somewhat humorous. 
     The parent's apparent confusion was simply another indication of how ineffective traditional instruction in grammar has been. It called to mind the contributor to NCTE-Talk who insisted that we should be teaching "transient" and "intransient" verbs. My reply to the parent was basically the same as the theme of this book. Forget the rules, forget the grammar books. Learn how to analyze sentences, and then look at some real texts. Do your own first-hand research. It won't take very long to find acclaimed writers who begin sentences with "But" and/or with prepositional phrases.
      The problem is that we don't teach students to identify basic grammatical constructions in real texts, and thus we can't teach them to use their own eyes and brains to make their own judgments based on their own primary research. There is an interesting story about Francis Bacon, one of the major figures in the scientific revolution. When he went to college, Aristotle and the Scholastics reigned supreme. It was literally forbidden to challenge Aristotle. Bacon, of course, is generally acknowledged as the father, or one of the fathers, of modern scientific experimentation -- look at the real world (not at other people's conclusions about it), make observations, and form conclusions. I note the story because, in spite of all our professional claims to research, particularly in the area of grammar, and in spite of Martha Kolln's "Closing the Books on Alchemy," (College Composition and Communication, 1981, Vol 32, 139-151) we are still, now in the 21st century, alchemists or scholastics when it comes to grammar.  We have, of course, deceived ourselves. We claim to be scientific. We venerate "the research"; but rarely do we actually read it, or better yet, do it. Instead, we parrot its conclusions, particularly if those conclusions support our own prejudices. But "research" does not make us scientific, for "scientific" describes a state of mind, a state of mind that looks, not primarily at the results of someone else's "research," but rather at the real world around us, in our case, living texts.
     There is, we need to note, a good reason for our failure. As a profession, we have not prepared ourselves to do so. I don't know whether or not the parent who wrote to me can identify prepositional phrases, but I do know that many teachers of English cannot. And many cannot identify subjects, verbs, clauses, etc. We have not taught them how to -- and they have not been encouraged to learn how to do so by themselves. Instead of teaching them a few basic constructions and then having them use those constructions to analyze texts (such as samples of what their future students will write), we have sent the teachers to the grammar books, books which are now more confusing and complicated than they ever were before. Our future teachers can ace these courses, and still go into their new classrooms totally unable to identify the phrases, subjects and verbs, clauses, etc. in what their students write. Likewise, they cannot intelligently read the research, and thus they have no choice but to accept its conclusions. The fault, of course, is primarily ours, not theirs. But we need to keep this situation in mind as we look at the often acclaimed, but poorly understood, research about the teaching of grammar.

     We need to realize that the anti-grammar "the research proves" argument has become a myth -- justifying the prejudice (and perhaps ignorance) of people who don't want to teach it. Its strength lies, not in the research, but in the constant repetition of the conclusion.  I'll never forget lunch at a Delaware Valley Writing Conference, many years ago. I got my lunch and sat down with someone whom I did not know. We started to chat, and he asked if I was presenting a paper. When I said "Yes," he asked what it was about. When I told him "grammar," he asked why I would present a paper about teaching grammar when "all of the research" demonstrates that teaching grammar is useless or even harmful. I asked him to name one of the research studies. He mentioned O'Hare's. I told him what was wrong with O'Hare's study (See below.), and asked him to name another. He did so, and I gave him a critique of that one. If I remember correctly, he was able to name a third, the flaws in which I also pointed out. As he got up to leave, he looked back and said, "Well, somewhere out there, there are studies that prove that teaching grammar is harmful."  His attitude, unfortunately, is still shared by too many members of our profession.  Fortunately, the majority within our profession is more open-minded. The following discussions, therefore, attempt to explain what makes sense, and what doesn't, in some of the more influential and more interesting research studies.

1963 The Braddock Report (and Martha Kolln's Response)

     Research in Written Composition (NCTE, 1963), written by Richard Braddock, Richard Lloyd-Jones, and Lowell Schoer, contains the statement that is most often quoted to deride the teaching of grammar:

          In view of the widespread agreement of research studies based upon many types of students and teachers, the conclusion can be stated in strong and unqualified terms: the teaching of formal grammar has a negligible or, because it usually displaces some instruction and practice in actual composition, even a harmful effect on the improvement of writing. (37-38)
In "Closing the Books on Alchemy," Martha Kolln explores several of the problems with this conclusion, but she misses some, and others need to be emphasized.
     Kolln notes, for example, the authors' tentative terms, their reservations, throughout most of the book -- with the exception of the section on grammar (140). Perhaps because she did not have the space, she does not address the purpose and scope of the book itself. On the first page of the book, the authors report that it was written at the request of the Executive Committee of the NCTE:
the Executive Committee appointed an ad hoc Committee on the State of Knowledge about Composition "to review what is known and what is not known about the teaching and learning of composition and the conditions under which it is taught, for the purpose of preparing for publication a special scientifically based report on what is known in this area." (1)
We are then told that
The committee further decided to use only research employing "scientific methods," like controlled experimentation and textual analysis. At the suggestion of the Executive Committee, the ad hoc committee set as its goal the identification of the dozen or so most soundly based studies of the foregoing type. (Actually, the committee finally identified five such studies, each of which is summarize in detail in Chapter IV.) (1)
If we read the Braddock report for what it claims to be, it is, with one major exception, a very fine piece of work. But what it claims to be is an exploration of the state of knowledge (in 1963) -- and methods for adding to that knowledge. Most of the book concerns how to design research projects, and the commission selected for emphasis the best designed projects that they could find. They were hoping for a dozen; they found five. (They started with over 1,000 bibliographic entries, narrowed that to 485 which were then further screened, and ultimately settled on five.)  Later, we are told that "the writers of this report were tempted more than once not to select any 'most soundly based studies' to summarize at length . . . ." (55) As they themselves noted, "Composition research ... is not highly developed."
     What we have in the Braddock report, then, is research on doing research. The authors were appointed. None of them claim any expertise, either in grammar, or in the teaching of it. They based their "harmful effects" statement primarily on one of the five studies (done by Roland J. Harris), and that study was chosen because of the methods used in it. None of the authors of the report appears to have any qualifications to judge the intrinsic, as opposed to methodological, validity of Harris's work (70 - 83). And if we look at the methodological aspect of Harris's study, we can see why the authors of the Braddock report found it, if not good, at least better than most. Harris's study covered a two-year period, far better than the typical one-year study. Harris conducted a pilot study. He fairly clearly describes the differences between the "Formal Grammar" and "Direct Method" groups. (Kolln is clearly right in noting, as do the authors, that both groups were taught grammar.) He describes his procedures, including the procedures for selecting students, and he gives a fairly comprehensible and complete account of the results. 
Methodologically, this may, in 1963, have been a model study.
     But what about the intrinsic value of the study? In view of the NCTE resolution against the teaching of grammar not supported by theory and research, we need to note immediately that Harris provides, at least as reported by the authors, absolutely no theory. He has no theory of natural language development, and he has no "theory" regarding the causes of "errors." This omission raises serious questions about all the variables that he counted. He counts, for example, the "omission of a full stop" (apparently comma-splices and run-ons) as an error without seeming to realize that such "errors" are 1) often a sign of growth, and 2) often a sign of poor or incomplete instruction in grammar. They often occur, for example, when writers sense a close relationship between two main clauses, but do not know how to punctuate them. Likewise, he counts the "misuse of object forms of pronouns" as an error, but there are big differences between  "To who did you give it?" and "Who did you want to see?" Without a closer examination of these errors, the results, in other words, are meaningless. What impressed the authors of the Braddock report, in other words, were the way in which the study was conducted and the way in which the results were presented. They did not look at the validity of the variables that were being counted, even though serious questions can be raised about each of the eleven of them.
     Of more interest here, however, are Harris's conclusions and the differences between his control and experimental groups. The Braddock report tells us that "the Formal Grammar group followed a logically organized program of traditional grammar instruction 'through the parts of speech, with stress on the function of words' and employed the traditional grammatical terminology in classroom teaching and in correcting compositions." Teachers in the "Direct Method" sections were apparently instructed not to use grammatical terms, although there is no way of knowing whether or not they did. What Kolln misses is that the "Formal Grammar" group may have been taught grammatical terminology, but they did not learn it. Both groups were given a test of their knowledge of formal grammar (described as "Test C"). According to Braddock et al., "The investigator also expressed dissatisfaction with the low levels of achievement of the Formal Grammar group on Test C, only one class having more than 50 percent of the answers correct." (83) Only one class, out of five, scored better than 50% on the final grammar test!
     Perhaps because they wanted to present it as a model study, the Braddock report does not explore many of the other weaknesses in the Harris study. It does, however, indirectly suggest another one. In a section called "Unexplored Territory," they suggest further research on twenty-four questions, the third of which is "What are the sources of fear and resentment of writing?" (52) Most people don't need research to realize that one of the main sources of fear is the fear of making grammatical mistakes. We cannot tell the extent to which it was done, but it is clear that in the Harris study, the Formal Grammar group received grammatical instruction "in correcting compositions." For them, in other words, "writing" meant applying those less-than-half-learned grammatical rules. Would anyone want to argue that this probably did not have an effect on their writing in the post-tests?
     In fairness to Harris, it should be noted that his study was done as a doctoral dissertation -- which, in itself, is something that the Braddock report derides. (33) Harris's own conclusion, however, is remarkably similar to, yet more judicious than the Braddock report's "harmful effects" statement:
It seems safe to infer that the study of English grammatical terminology had a negligible or even a relatively harmful effect upon the correctness of children's writing in the early part of the five Secondary Schools. (83)
There is a major difference between "the study of English grammatical terminology" and "the teaching of formal grammar." Indeed, why would anyone assume that the teaching of grammatical terminology, in itself, would improve a person's writing, especially when that terminology is less than half learned and often does not directly apply to writing? Perhaps the only major flaw in the "harmful effects" statement is the failure to include the word "terminology." Its inclusion might have shifted the debate from the fruitless one (Should "formal" grammar be taught?) to a more productive one (Can grammar be taught with no, or at least less, focus on terms?)
     Kolln, like many advocates of the teaching of grammar, confuses this central issue. She argues that even though Harris's experimental group did not use terminology, they did study grammar. Likewise, she argues that the students in the experimental groups in O'Hare's well-known Sentence Combining: Improving Student Writing without Formal Grammar Instruction used terminology, albeit significantly different from the traditional. ("That they labeled clause modifiers as 'who statements' rather than 'relative clauses' doesn't mean they didn't study grammar." (149) She then observes, "We have been warned so often against teaching grammar for its own sake, we are even wary about asking our students  to learn basic terminology." The problem is that she never defines "basic." Instead she concludes "Our goal should be to help [students] understand consciously the system they know subconsciously as native speakers, to teach them the necessary categories and labels that will enable them to think about and talk about their language, so that when they use it consciously, as they do in writing, they will do so with control and grace and enthusiasm." (150) Basically, in other words, Kolln uses research about teaching grammar without terminology, or with minimal and significantly different terminology, to justify teaching grammatical terminology.
     The question is, if we really want, as Professor Kolln suggests, "to teach [students] the necessary categories and labels" is there a better way to go about it? This book is, of course, an attempt to show that there is.  The details of the KISS Approach are discussed later in this book. The point here is that the approach differs significantly, both from most of the current grammar textbooks, and, of course, from the methods used by the "Formal Grammar" group in the Harris study (and all the other studies which show that focussing on terminology does not improve writing). In the Harris study the students were apparently taught a ton of grammatical terms, over half of which they did not learn. In a KISS approach, students are introduced to a few terms at a time, and they work with those terms until they have mastered them. Then they move to the next level and add a few more terms. The conclusions of the Harris study, and the misstated "harmful effects" statement of the Braddock report, simply do not apply to a KISS approach.

1966 The Bateman and Zidonis Study

     In 1966, NCTE published The Effect of a Study of Transformational Grammar on the Writing of Ninth and Tenth Graders, by Donald Bateman and Frank Zidonis. In debates about the effectiveness of teaching grammar, this study has been frequently cited as proving the ineffectiveness of teaching grammar, but the study "proves" just the opposite. Bateman and Zidonis note that their results are tentative. "Even so, the persistently higher gain scores for the experimental class in every comparison made strengthens the contention that the study of a systematic grammar which is a theoretical model of the process of sentence production is the logical way to modify the process itself." (37) They further note that "the persistent tendency of researchers to conclude that a knowledge of grammar has no significant effect on language skills (when judgment should have been suspended) should certainly be reexamined." (37)

1969 John Mellon's Experiment

     In 1969, NCTE published John C. Mellon's Transformational Sentence-Combining: A Method for Enhancing the Development of Syntactic Fluency in English Composition. The 247 students in the experiment were divided into three groups. The experimental group did sentence-combining problems; the control group did "traditional parsing exercises"; and the placebo group did no grammar. (31) He reported that "significant growth of syntactic fluency occurred in the writing of the experimental group." (52) But in reporting on an overall quality comparison of the writing of the three groups in the study, Mellon states: "the writing of the experimental group was inferior to that of the subjects who had studied conventional grammar, but indistinguishable from that of subjects who had studied no grammar but had received extra instruction in composition." (69) In that one of the objectives of sentence combining is to produce longer main clauses, the length and convoluted (deceptive?) nature of Mellon's statement is particularly interesting. Put more simply, it means "The traditional grammar group wrote better than either the experimental or the placebo groups." Mellon adds "curious results indeed."
     Throughout his report, Mellon makes several interesting comments about the teaching of traditional grammar. For example:

it may very well be the case that conventional grammar study fails to promote growth of syntactic fluency not because of the usage practice which it features, but rather because of the hundreds of simply structured and altogether childish sentences which it employs for parsing exercises. (63)
The logical conclusion to Mellon's study is that, if we want both increased fluency and quality, conventional instruction should be modified to include more mature sentences. This conclusion, however, was never developed. (Note that a primary objective of the KISS Approach is to enable students to analyze and discuss the structure of any sentence that they read or write.)

1971 The O'Hare Study

     In part Mellon's ideas were not developed because of Frank O'Hare's Sentence Combining: Improving Student Writing without Formal Grammar Instruction, published by NCTE in 1971. What O'Hare advocated is pure exercise in sentence-combining, with no instruction in grammar. Students, in other words, are not to be told what constructions they are manipulating. As for why they are manipulating them, they must rely on the instructor's assertion that it will help them write "better." O'Hare's approach to sentence combining is thus pure operant conditioning. Few people would argue that operant conditioning is not effective, but there are some serious questions about its long-term effects. A general principle of operant conditioning is that once the stimulus-response treatment has ended, the bond between the stimulus and response is gradually, if not quickly, worn down by competing stimuli. O'Hare's experiment lasted eight months, and his results are based on pre- and post-treatment writing samples. If we are to be convinced that the treatment has long-term effects, should there not have been another post-test, perhaps even six months later? Smith and Hull (RTE, Oct 83) noted a significant decrease in the effect of sentence combining one week after instruction ended! O'Hare, however, never addresses this question. Since O'Hare bases his conclusions on both an assessment of syntactic maturity and on an assessment of writing quality, let me briefly examine each.

     The Assessment of Syntactic Maturity

     What was O'Hare measuring? In effect, he measured two things: words per main clause (T-unit), and subordinate clauses per main clause. These two things were divided into six categories: "words per T-unit," "clauses per T-unit," "words per clause," "Noun Clauses per 100 T-units," "adverb clauses per 100 T-units," and "adjective clauses per 100 T-units." He justifies these measures by referring to the work of Hunt and O'Donnell. Discussing Hunt's 1970 study, Syntactic Maturity in Schoolchildren and Adults, O'Hare writes: "Although Hunt used a number of new measures, and got especially interesting results with what he called structures less than a predicate and less than a clause, his findings relating to the number of embedding transformations and to clause and T-unit length were of particular interest to the present study." (24) O'Hare intentionally downplays, in other words, Hunt's statement that "the great majority of the syntactic changes that increase with maturity are those that reduce a clause to less than a clause." (43) In his study, Hunt examines things such as verbals and appositives, constructions that reduce not only the length of a T-unit but also the number of subordinate clauses per T-unit. Such constructions, says Hunt, are most indicative of syntactic maturity. Why did O'Hare ignore this?
     The importance of reduction to syntactic maturity is noted by both Hunt and by O'Donnell, who even goes so far as to question the validity of counting subordinate clauses as a measure of syntactic growth. (98) (O'Hare doesn't mention this.) The following example suggests the affect of reduction on a statistical analysis:
     a.) They bought a house which was new and expensive.
     b.) They bought a new, expensive house.
Most readers would consider (b) as a better stylistic norm than (a), but, on the whole, (a) gets a better rating on O'Hare's measures: (a) has 9 words per main clause to (b)'s 6, and it has 50 adjective clauses per 100 main clauses, to (b)'s 0. (b) does get a better rating on words per clause (6 to 4.5), but this difference raises another question about O'Hare's study. If we analyze the results of the experimental group (54), we find that there was a 63.6% increase in words per main clause, but only a 21.1% increase in words per clause. May this not indicate that these seventh graders used subordinate clauses where twelfth graders would have used reduced constructions?
     By themselves, clause length and subordinate clause ratios are insufficient measures of syntactic maturity. Much depends, as Hunt suggests in his 1970 study, on the content of the clauses. And, as O'Hare himself suggests in his conclusions: "There is need for massive research into what constitutes a mature style." (77) In Classical Rhetoric for the Modern Student, for example, Corbett suggests counting the number of sentences that are 5 words below average and 10 words above. Analyzing the 35 passages from well-known writers that Corbett provides as models, I found that 18% of the main clauses were 10 or more words above the average (which was 22.9), and 47% were 5 or more words below the average. A comparable analysis of the writing of college Freshmen indicated 7% that were ten or more above the average (15.3), and only 24% that were 5 or more below. Perhaps the variety of main clause length is as significant as the length itself.
     If we turn from what O'Hare was counting to the subjects of the experiment, we find still another reason to question his counting subordinate clauses. Early in his report, O'Hare includes a chart that combines the results of the studies of O'Donnell and Hunt. The numbers in O'Hare's chart for clauses/T-unit are those in the left column below:

Grade Clauses / T-unit Difference Avg. Difference / year
3 1.18
4 1.29 .11 .11
5 1.27 -.02 -.02
7 1.30 .03 .015
8 1.42 .12 .12
12 1.68 .26 .065
Adult 1.74 .06

What these statistics suggest, even though they may not be very sophisticated, is that seventh grade is the period of the largest natural increase in subordinate clauses per main clause. Although this may seem a trivial objection, it is not, given the fact that O'Hare does not account for reductions. It is commonly acknowledged that children, in mastering a new construction, overgeneralize (as in the youngster's "I readed a book"). The .12 increase from seventh to eighth grade in the preceding table, since it is the largest average yearly difference, may already reflect this overgeneralization. (Note, for example, that the .11 increase from third to fourth is followed by a .02 decrease: if we had the statistics for ninth, might not they too show a decrease?) What O'Hare may have done, therefore, is not accelerate "normal" growth, but rather aggravate an abnormality. This is also the place to point out that O'Hare is guilty of what Piaget calls the American fallacy -- the attempt to accelerate normal growth rather than understand and guide its direction. As he says, "the control group showed only 'normal' growth." (70)
     Still another questionable aspect of O'Hare's statistics involves his comparison to Mellon. He claims to be studying seventh grade students because Mellon did and he intends to compare his results to Mellon's. Likewise, he adapts Mellon's exercises in an attempt to keep the studies comparable. Yet, when he comes to his method of counting constructions, he writes:

          Mellon counted clauses of condition, concession, reason, and purpose as separate T-units because he believed that logical conjunctions behave much like coordinate conjunctions. In addition, he discarded clauses with repeating predicate phrases because he claimed they were elliptical and therefore vacuous. This experimenter remained unconvinced by Mellon's reasoning in either case and, therefore, retained Hunt and O'Donnell's simpler and more convincing methodology. (48)
Mellon does not give an example of such a clause, but he does list examples of the conjunctions, one of which is "because." (43) What effect might this change in procedure have on the statistics? Suppose, in the pretest, a student wrote:

I'm sure glad tomorrow is Friday. I just love weekends.

but on the post-test:

I'm sure glad tomorrow is Friday, because I just love weekends.

The pre-test sentence would receive the same rating from each researcher: 5.0 words/T-unit, 3.7/clause, and 50 subordinate clauses per 100 T-units. On the post-test sentence, however, the ratings would be significantly different: Mellon would arrive at 5.5 words/T-unit, 3.7/clause, and 50 subordinate clauses per 100 T-units, essentially the same as on the pre-test. But O'Hare's post-test results would be 11 words/T-unit, 3.7/clause, and 200 subordinate clauses per 100 T-units, i.e., twice the words per T-unit, and four times as many subordinate clauses! Yet, when he comes to comparing his results with Mellon's, he simply states: "The experimental group's mean pre-post change score of 6.12 words per T-unit was approximately five times the statistically significant increase reported for Mellon's experimental group." (55) He makes a similar statement two pages later, but does he even suggest that his way of counting may have influenced his results?

     The Assessment of Writing Quality

     Since he edited and corrected the students' papers before he gave them to evaluators, O'Hare has a limited definition of "writing quality." As he says:

This study was interested in the students' writing ability and not at all in their spelling, punctuation, or handwriting talents. In order to eliminate the possible effects of these extraneous factors on the evaluators' judgments, the thirty pairs of compositions were typewritten so that spelling and punctuation could be corrected. (51)
In her response to Patrick Hartwell's use of O'Hare's study, Carole Moses observes, "I wonder how many fragments and run-ons were corrected." (CE 47 Oct 85, 646.) I would add comma-splices, and the various misspellings of "to," "have" (i.e., "of"), "it's," "their," etc. The question is particularly interesting since Hunt, as early as 1965, wrote: "As more nonclausal structures are packed into a clause the likelihood of stylistic faults occurring increases apace. The greater the congestion the greater the hazard" (152). Having analyzed a couple hundred passages, I find it difficult to believe that O'Hare could not have counted the fragments, comma-splices, and run-ons per main clause. Apparently, had he done so, he could not then have made the claim that his approach "improves" students' writing.
     Having noted the "notorious unreliability of composition ratings," O'Hare goes on to tell us that his raters were "eight experienced English teachers who were attending Florida State University during the summer of 1970." (50) Although he states that "these evaluators had no knowledge of the nature of the present experiment," I would like to know what courses they were taking. Just as O'Hare assumed that there would be a "rubb-off" effect from the sentence combining exercises to the students' writing, so there could be a "rubb-off" effect from a course in transformational grammar, advanced stylistics, or language development, should one or more of these teachers have been taking it. This possibility is not that far-fetched, since we are told that the teachers "were simply told to make a single judgment on the overall quality of the compositions in each pair, basing their decision on ideas, organization, style, vocabulary, and sentence structure." (50) Doesn't "style" include "sentence structure"? (If it doesn't, what does it mean?) O'Hare notes, moreover, that the evaluators had a "rater-training" session, without which, such evaluations have been shown to be meaningless. Since the errors in the sentences had been corrected, what was said in that session about "style" and "sentence structure"?
     O'Hare goes to great length, defending his choice of the single judgment over a rating scale (i.e., 1 to 5). Did he not think of, or did he intentionally avoid, another possibility: having the raters make a single judgment about each pair for each criterion. We then would have been able to tell whether the essays were being rated for ideas, organization, style, etc. As the system stands, however, some of the "winners" may have been chosen for ideas, others for organization, etc. This problem is particularly bad since only post-test essays were evaluated! The papers were paired for sex and IQ, but these factors do not account for creativity, use of detail, etc. Thus O'Hare later notes that the experimental group's essays "had much more detail, more 'meat' to them." (72). How do we know that the pre-test essays of this group were not also superior in detail? Finally, we should note that only narrative and descriptive essays were evaluated. As most English teachers agree, when students write expository essays, their organization, vocabulary, sentence structure, etc. often falls apart. It would be nice to know what the judgment would have been on expository writing.
     In effect, all O'Hare has proven is that sentence combining increases clause length and the use of subordinate clauses during the course of instruction. But as I suggested above, this may be an aggravation of an abnormality rather than an "improvement." As Hunt wrote in 1965: "In this study the word 'maturity' is intended to designate nothing more than 'the observed characteristics of writers in an older grade.' It has nothing to do with whether older students write 'better' in any general stylistic sense." (5) Having looked at O'Hare's study in some detail, might we not ask how he can possibly state that he is "Improving Student Writing"?
     Perhaps the most destructive critique of O'Hare's study comes from O'Hare himself. In 1986 he published The Modern Writer's Handbook (NY: Macmillan), the first half of which is a very traditional presentation of grammar. Apparently, O'Hare himself no longer believed in the validity of his anti-grammar study.

1979 Elley, Barham, and Wyllie

     The last of the most influential studies that have been used to "prove" that instruction in grammar is useless is The Role of Grammar in a Secondary School Curriculum by W.B. Elley, I.H. Barham, H. Lamb and M. Wyllie, published by the New Zealand Council for Educational Research in 1979. What this study proves, however, is that formal study of transformational grammar, as presented in the Oregon Curriculum, is not effective. But the Oregon Curriculum does not teach students how to use transformational grammar to analyze their own writing. Like so many of the traditional grammar texts, the Oregon Curriculum focuses on the complicated rules of transformational grammar. To assume that such instruction would improve students' writing is comparable to assuming that a detailed course in human anatomy would improve a student's chances of catching a football. Anti-grammar zealots among us, however, quote the conclusions of this study as if they apply to any systematic approach to teaching grammar.

1986 The Hillocks Report

      In 1986, ERIC and NCTE published Research on Written Composition: New Directions for Teaching, by George Hillocks, Jr.  In the "either you're for it or against it" world of the great grammar debates, the Hillocks report has been used as another torch in the inquisition against grammar. But the Hillocks report, in this debate at least, is deceptive. The question is not that simple. Aware, of course, of sentence-combining, Hillocks treats it as close to, but separate from, grammar. In the section on grammar, he basically repeats the conclusion of the Braddock report:

 None of the studies reviewed for the present report provides any support for teaching grammar as a means of improving composition skills. If schools insist upon teaching the identification of parts of speech, the parsing or diagramming of sentences, or other concepts of traditional school grammar (as many still do), they cannot defend it as a means of improving the quality of writing. (138)
This is a very unfortunate and inaccurate statement indeed, especially since Hillocks later praises, and praises very highly, a study which did indeed included the identification and study of grammatical constructions. The difference, however, is that Faigley's study employed a KISS, rather than a Traditional/Formal approach to teaching grammar.
      In his discussion of Sentence Construction, Hillocks gives high praise to a 1979 study by Lester Faigley: "The most carefully designed and implemented study of sentence construction techniques is Faigley's . . . ." (146) He praises Faigley's study in a number of ways, and Faigley's study is one of the models for "New Directions." Unfortunately, however, he did not study Faigley's study as well as he should have. Later in the book he states "Sentence combining activities do not focus on the identification of parts of speech or parts of sentences but on the manipulation of syntactic elements, and, in the case of Faigley (1979c), on the generating of the elements as well." (214) Perhaps the key word here is "focus," but in order to understand what went on, we need to look at Faigley's study.
     Perhaps because of the anti-grammar political pressure, Faigley was not very clear about what the experimental group in his study did. In one of his articles on the study, he states that "Neither group was taught formal grammar, and the experimental staff employed ad hoc terminology, such as '-ing phrase,' when referring to a particular construction." (199) ["The Influence of Generative Rhetoric on the Syntactic Maturity and Writing Effectiveness of College Freshmen," Research in the Teaching of English, Vol. 13, 1979, 197-206.] But in the preceding paragraph, he stated, "The experimental sections followed the arrangement of Christensen and Christensen's A New Rhetoric, writing in class many of the exercises in the text." To see the problem here, we need to look at the Christensen's A New Rhetoric.
     A New Rhetoric consists of two parts -- "I The Process of Writing," and "II The Larger Units of Composition."  In spite of its title, Part I is filled with  (but with a limited number of) grammatical terms. Opponents of instruction in grammar would be horrified to read:
It is hardly necessary to insist again that the meaning, or perhaps here the interest, is in the modifiers. We need now a language for discussing them. The next few paragraphs are the foundation of our treatment of the sentence. You should master them so well that you can apply the principles creatively, in writing, and analytically, in the discussion of writing. The language we need is of two sorts--grammatical and what we will call rhetorical." (22, my emphasis) 
The paragraphs that follow this statement assume knowledge of "subject,"  "verb," and "subordinate" and "main" clauses. They explain, among others, adverbs, prepositional phrases, verbals, verb phrases, absolutes, relative clauses, adverbial phrases, adjectival phrases, "free noun phrases" (which turn out to be, and be called, appositives), and "the noun with an adverbial function...".
     Students who used the Christensen's book did not study formal grammar in the traditional approach, but they very clearly studied grammar, including the ability to name and identify specific grammatical constructions, in the sense of a KISS approach. This is clear, not only from the numerous examples and exercises, but also from the basic structure of Part I. Because the book is for college students, the Christensens are able to start early with the concept of the "Two-Level Narrative Sentence." In a "two-level" sentence, the first level is the base, or main clause. The second level is subordinated modification, in the form of various grammatical constructions. The bulk of Part I of the book consists of examples of sentence levels, with the grammatical constructions used to create the subordinated levels indicated in bold. For example (from page 47):
1. And the eyes were calm,
     2. aware but not interested. (Adjective Plus Adjective)
Page after page, a dozen or so of such examples is preceded by a short explanation of the stylistic effects of different variations. Then a new construction or variation in levels is explained and more examples are given. Having studied the examples, the students are expected to do exercises in which they identify the levels and constructions. Once these have been discussed, the students are expected to start with a base sentence and generate such sentences themselves. If the students in Faigley's study used the text, as he said they did, they could not possibly have avoided the instruction in grammatical terms and concepts. 
     As did several other researchers, Faigley attempted an evaluation of overall quality, an evaluation in which his experimental group was judged to have written better papers. He also followed through with some statistical procedures in an attempt to determine what it was about the writing of the experimental group that accounted for the difference. As Hillocks puts it, "Stepwise regression showed length and percent of T-units with final free modifiers to be the two most important variables associated with quality, a result corroborated by Nold and Freedman (1977), who found that length and the percent of words in final free modifiers were the most important variables associated with quality." (147) This raises the question, however, of whose sense of "quality," (I find it rather hilarious that, as I quote these sentences from the researchers, my grammar checker regularly underlines them as "long sentence.") Hillocks goes on, however, to quote Faigley's suggestion that the approach works "by stressing the addition of specific details to abstract statements as a means of generating content." (147) The comment deserves some elaboration.
     Most writing teachers would love to be able to get more of their students to include more relevant details in their writing. "Details," however, is a very abstract word, and the suggestion "Give more details" usually doesn't work. In the Christensen's approach, however, the students start by learning how to identify several grammatical constructions, and they are given numerous examples of how those constructions add details to the writing. Then the students are trained in adding these constructions to a base sentence. From the students' perspective, I would suggest, those directions are much more concrete. "Add details," if understood in the first place, could mean to add additional sentences. "Add a prepositional phrase," or "Add an appositive," are not only much more specific (if the student understands what these are), but they also result in the details being added within an already written sentence.
      One has to wonder why, in light of the praise given to Faigley's study by Hillocks, the Christensens approach has not been pursued. Part of the problem is probably Faigley's downplaying, for whatever reason, of the role of grammar. Part of the reason is Hillocks' misreading of (or insufficient research on?) the study. Instead of showing how the Christensens take a different approach to teaching grammar, Hillocks, in effect, turned Faigley's study into an argument against teaching any grammar at all! As a result, anyone who looked at the Christensens' book would have been prejudiced against its heavy emphasis on grammar. Add to that the fact that many teachers have themselves not been taught how to identify appositives and absolutes. For them, to use the book would mean that they themselves would have to work their way through it (without the help of an instructor), just to be able to understand it. Why should they do that when "all the research shows ."?

      Opponents of teaching grammar (either at all, or systematically) will, as suggested near the beginning of this chapter, forever go on proclaiming that "somewhere" there is research. It is, however, impossible to discuss all of the studies here. Debate about such studies is welcome on the this web site, but real debate about the studies seems to be anathema to those opposed to grammar. Debate requires an exploration of the details, and, of course, the devil is in the details. And, as the preceding detailed discussion of Faigley's study indicated, looking at the details might reveal that a study which has been used to condemn the teaching of grammar might actually support it. Faigley's study, of course, does not support the "Traditional Formal" approach to teaching grammar, but it does support the fundamental KISS approach. I have often been asked if there is any research that supports the KISS Approach, and, until I reviewed the research in preparation for this chapter, I had always pointed out that such research is expensive. Now, I can point to Faigley' study, but I can also claim at least three, possibly four additional studies, all published in a relatively little-known book.
     The book, edited by, Donald A Daiker., Andrew Kerek, & Max Morenberg, is called Sentence Combining and the Teaching of Writing: Selected Papers from the Miami University Conference, Oxford, Ohio, October 27 & 28, 1978. (It was published by The Departments of English, University of Akron and the University of Central Arkansas, 1979.) As its title suggests, it consists of selected papers from a conference. What the title does not suggest is that the conference was, in effect, a victory celebration for advocates of sentence combining. It included presentations by many of the big names associated with the movement -- John Mellon, James  Kinneavy, Kellogg Hunt, Donald  Daiker, Andrew Kerek, Max Morenberg, and William Strong. The celebrational tone of the conference is evident throughout the papers, both in terms of the celebrities presenting and in terms of "We won." What's interesting, however, is the "pro-grammar" undercurrent in several of the papers, if not in the conference as a whole.
      Teachers who need to have "research" to justify the teaching of grammar really should read this book -- it is impossible to summarize the nuances, statements, counterstatements, etc. expressed in its 228 pages. It is interesting, for example, to note that O'Hare, the most well-known advocate of "pure" sentence combining, apparently did not present a paper at the conference. (As noted above, he later wrote a very traditional/formal handbook.) Hunt's presentation was basically a personal reminiscence, but among other things he noted:

I remember the promise that transformational grammar had for language education fifteen years ago. Then extravagant promises were made. Its rise was meteoric, and the shower of sparks from its tail was golden. But when that sun exploded, it left a black hole where grammar used to be. And just as there are black holes out in space where no matter can enter, so no grammar-like study will be tolerated in the schools again for many years to come. (153)
"For many years to come" is not "forever." Reading Hunt's statement, we might even say that he prophesied, and even looked forward to, the return to the curriculum that grammar has been making in the last several years. Hunt's presentation was not about a research study. His major work never concerned the "horse-race" experiments that pitted one instructional approach against another. 
     Of the four research reports I want to look at, the first, surprisingly for this conference, presents a problem with sentence combining. As noted previously, all of the major studies basically ignored errors.  In "Words Enough and Time: Syntax and Error One Year After" (101-108), Elaine P. Maimon and Barbara F. Nodine explored the possible increases in errors as a result of sentence-combining practice. Although their presentation, necessarily short, does not make it clear, they apparently followed O'Hare and Strong (See page 102) and used exercises without grammatical terms. They looked at faulty subject-verb agreement, dangling verbals, misplaced modifiers, fragments, vague pronoun reference, faulty parallelism, and comma-splice/run-on sentences. They used Hunt's "Aluminum" passage for both the pre- and the post-test. Embedded errors per 100 words increased from .48 on the pre-test to 1.25 on the post-test. In other words, the rate of errors almost tripled! Earlier in the conference, Mellon had stated "Sentence combining produces no negative effects . . . ." (35) It would have been interesting to know his response to the paper by Maimon and Nodine.
     The preceding study is the one that I consider "possible" support for the KISS Approach. It suggests that pure sentence combining exercises may result in more harm than help. The next study is by Rosemary Hake and Joseph M. Williams. Their "Sentence Expanding: Not Can, or How, but When" (134-146) suggests that attaining writing competency may be related to reducing [not increasing] word/T-unit and clause/T-unit counts. It also supports the KISS ideal curriculum:
     Students whose pre- and posttest essays were judged competent significantly increased their word/T-unit and word/clause counts, but not their clause/T-unit counts.
     Students whose pre- and posttest essays were judged incompetent did not significantly change their T-unit counts, counts that still remained higher than the essays judged competent.
     But most important were the counts of the students whose pretest essays were judged incompetent but whose posttest essays were judged to be competent. They had significantly decreased their word/T-unit and clause/T-unit counts.
     It was at this point that the issues became more fundamental than simply the relative effectiveness of combining and imitation exercises. The different responses to these expansion exercises by students at different levels of competency suggest that the exercises' usefulness may vary according to the abilities of a student. We therefore began to ask a different question: not just how sentence combining should be taught, but when. One answer suggested by the above data seems to be only when a student is ready for it, only when he is already a competent writer or ready to become one." (139)
If I am reading this correctly, it confirms something I have long expected. Sentence-combining exercises, basically divorced from instruction in grammar, may help the better students to become still better, but they just waste the time of the weaker students in the class. Is this what we are supposed to be doing?
     The idea that sentence-combining exercises help students only when they are ready for them actually supports the ideal KISS Curriculum. The sentence-combining texts that I have seen have been developed with no sense or theory of natural syntactic development. Students at all grade levels, and at all levels of proficiency, are expected to create combinations using all of the possible grammatical constructions.  When my son was in second grade, one of his sentence-combining exercises involved combining "Mary is a biologist" and "Mary studies fish." He would not, however, accept "Mary, a biologist, studies fish" as an acceptable sentence, even after I did the exercise for him. As will be discussed in the next chapter, Hunt suggested the appositive (needed here) is a "late-blooming" construction, possibly blooming as late as ninth or tenth grade. What I am suggesting, of course, is not that sentence-combining should not be used with younger students, but that the exercises should be based on a theory of natural syntactic development.
     Whereas the Hakes and Williams presentation supports only the sequential nature of the KISS Approach, Jeannette Harris and Lil Brannon argue that, at least at the advanced level, sentence combining is more effective if it is combined with instruction in grammatical terms. The combination enables the students to understand and discuss what they are doing and why: ("Sentence Analysis and Combining as a Means of Improving the Expository Style of Advanced College Students," 170 - 177)
     Working with ... students in the Writing Center at East Texas State University, we discovered that an approach which combined sentence analysis with sentence-combining practice yielded rapid and satisfying results in increased maturity and improved style of writing. While most proponents of sentence combining deplore the inclusion of any type of grammatical analysis, we found that with advanced students the addition of this component increased both the rate and quality of improvement. "  (170)
The grammatical terms that Harris and Brannon mention in this article are very close to the terms/concepts of the KISS Approach -- "subject," "adverb," "prepositional phrase," "participial phrase" (the KISS "gerundive"), "dependent adverbial clause," "noun clause," "infinitive phrase," "gerund," "nominative absolute," "direct object," "predicate adjective," "postponed subject," "coordinating conjunction." The students, using these terms, analyzed the syntax of their own writing and that of specialists in their field. When they did this, they saw for themselves the differences in syntactic style. According to Harris and Brannon, "Sentence analysis gives the students an objective means of looking at their own writing, and sentence combining gives them the means of improving it." (174) 
     It is interesting to note that Harris and Brannon consider right-branching less mature than left-branching constructions.  Chirstensen and Faigley, as noted previously, both pushed students toward right-branching. If teachers themselves disagree, do we have the right to push students in either direction? Unlike many of the other researchers, however, by teaching their students how to analyze sentences, Harris and Brannon respect their students' intelligence and give them the ability to choose for themselves.
      I have saved Anne Obenchain's study for last because it deals directly with middle and high school students.  In "Developing Paragraph Power through Sentence Combining," 123- 133) she explains a statistical study of sentence combining within the LINKS TO FORCEFUL WRITING program. Unlike most of the researchers, however, her approach clearly focuses on the relationship between meaning and syntax. Although she is mainly concerned with subordinate clauses, she argues for the main idea in the main clause, and also presents some compelling ideas for left-branching modification. (The latter is again delightful in view of the Christensens' obsession with right-branching.)  Perhaps because of the nature of the conference, she, like Faigley, downplays instruction in grammar in her approach: " we do not need to talk about any kind of clauses -- or about compound or complex sentences . . . . We need to talk only about supporting ideas and non-supporting ideas and the four kinds of connectives which are the LINKS TO FORCEFUL WRITING. " (126) It is interesting to note that the editors, claiming "space limitations," deleted "a lengthy excerpt of exercises and commentaries." (Ftn, 126) 
     If one looks at Part One (Sentence Power) of Links to Forceful Writing, one finds a text that is, in many ways, comparable to the Christensens'. Students are given numerous texts, usually short stories, and are asked to identify the coordinating and subordinating conjunctions (and adverbs) in them. They are then asked to respond to the text using specified grammatical constructions: "In this answer, use a CONJUNCTIVE ADVERB that shows RESULT to link the reason and the punishment, placing the conjunctive adverb wherever you find it most effective" (116). Note that the text explains both the grammatical constructions and the logical reasoning behind them. As with the Christensens' text, the 179 pages of Obenchain's are filled with examples, explanations of stylistic and logical implications, and exercises. 
     Readers who are already familiar with the KISS Approach should have no trouble in seeing that Obenchain was basically teaching KISS Level Three -- Clauses. She had a different way of classifying conjunctions ("four kinds"), but that is basically what she was teaching. And her statistical results demonstrated both increases in "maturity" (as defined, not by Hunt, but by the horse-racing researchers), and also a reduction in errors. (132) In effect, Obenchain may have some excellent materials for use with seventh, eighth, and ninth graders.
      It is possible to search for -- and find -- more studies that support the basic KISS Approach. But perhaps it is time to stop the nonsense. Perhaps the most interesting paper presented at the sentence combining celebrational conference was by Lester Faigley. His title reflects his thesis -- "Problems in Analyzing Maturity in College and Adult Writing" (94 - 100). A clearer presentation of his ideas, perhaps a development of them, appeared in College Composition and Communication -- "Names in Search of a Concept: Maturity, Fluency, Complexity, and Growth in Written Syntax." (Vol. 31, 291-299)  In his review of the four "concepts" named in his title, he shows problems with all of them. He also demonstrates that there are innumerable aspects of good writing that these concepts cannot measure. He therefore concludes, "Until researchers can devise measurements sensitive to these kinds of relationships, the notion of maturity in writing will remain a very elusive concept." (299)
     When looked at in detail, all of the research is open to very serious questions. We can, if we wish, let the people in our profession who hate teaching grammar (and/or who do not understand it) continue to tie us up in endless debates about questionable research.  Meanwhile, of course, grammar is back. We may agree or disagree with standardized testing, and we may agree or disagree about state and national standards. But the tests and the standards are being imposed. We will be teaching grammar, whether we want to or not. As a profession, however, we can affect what grammar will be taught, when it will be taught, and how. We have several options. We can continue the useless debates about "research" and the teaching of grammar. Or we can sit back and simply let the grammarians and publishers reimpose the traditional formal approach. Or we can reexamine our approach to teaching grammar, starting with our objective(s). If we want research to support us, we need better research that is embedded in a theory of natural syntactic development. And we need to better prepare future teachers to recognize basic grammatical constructions. Doing so will enable the teachers not only to understand and participate in the research, but also to be more effective in their classrooms.
     Judas went out and hanged himself. But we do not have to do likewise.

This border presents the left side of 
Evelyn De Morgan's
(British, 1850-1919)
The Storm Spirits
1900; The De Morgan Foundation, Battersea, London, England 
[for educational use only]
Click here for the directory of my backgrounds based on art.