June 26, 2010
The Printable KISS Workbooks Page Return to Background Essays
Ballet Dancers
in the Wings
(c. 1900)
by
Edgar Degas
(1834-1917)
Statistical Exercises
and KISS Grammar

      Although many English teachers are not enamored by statistics, statistical exercises are very important for two reasons. First, they can provide useful information about students' writing, not only to teachers, but also to the students themselves. Second, used within the KISS framework, they can be a primary source of motivation for students.
      Most of the research on natural syntactic development was based on statistical studies. In the 1960's, Kellogg Hunt demonstrated that the average length of students' main clauses (which he called "T-units") naturally increases with age. Hunt called them "T-Units" because of the lack of a standard definition for "main clause." Hunt's "T-unit" is the same as the KISS definition of a main clause. Before Hunt's work, researchers had been looking for a "yardstick" to measure "syntactic maturity"--the way and rate at which sentences naturally grow longer and more complex as people become more mature. Attempts to count words per sentence fail because third and fourth graders write long sentences by compounding main clauses, especially with "and." 
     Hunt's work was reinforced by the studies of Roy O'Donnell and of Walter Loban. In the following two tables, Loban's data was taken from Language Development: Kindergarten through Grade Twelve. Urbana, IL.: NCTE. 1976. 32. Hunt's and O'Donnell's data is from Frank O'Hare's Sentence Combining. Urbana, IL.: NCTE. 1971. p. 22.

Average Number of Words per Main Clause by Grade Level

Grade
Level
Loban's
Study
Hunt's
Study
O'Donnell's
Study
3 7.60   7.67
4 8.02 8.51  
5 8.76   9.34
6 9.04    
7 8.94   9.99
8 10.37 11.34  
9 10.05    
10 11.79    
11 10.69    
12 13.27 14.4  
Professional
Writers
  20.3  

The differences in the studies (such as O'Donnell's showing 9.99 words/main clause for 7th grade students and Loban's showing 8.94) should raise questions, but there is little doubt that the average number of words per main clause increases with age. Because a reader's brain dumps to long-term memory at the end of main clauses, the clearing of STM creates a rhythm to the text. Even if readers can not identify main clauses, they can surely sense the difference in rhythm.
     There are many questionable aspects to these studies. For example, what kind of writing did the student do? Narrative writing (stories), for example, almost certainly involve fewer cause/effect statements than do some expository topics. Then there are questions about the students' preparation on the topic that they were asked to write about. Perhaps most important, exactly how were the writing samples analyzed--what counted for what? Sometimes, for example, students' writing is illegible. How does one count what one cannot decipher? It was, I believe, Roy O'Donnell, who referred to these as "garbles." Many of these studies simply omitted garbles from the text. But how many garbles were there in the samples?
     I had the opportunity of meeting Roy O'Donnell at a national conference, and I asked him where the original samples were. His response was that they were probably in a box in someone's garage. It is an understandable response -- at that time, of course, there was no internet. If the samples of students' writing were scanned and put on the internet (which is now easily possible), such studies would be much more valid. Statistical studies, however, are typically expensive and very time-consuming, and few, if any, such studies have been done to follow up on this work after the seventies.  (That is why the KISS statistical studies section is called "Cobweb Corner.")
     All of these questions should make us cautious about how we use the results of such studies, but as general guidelines for what should be taught when, and as instructional exercises for students, these studies can be very helpful. 

     The studies that analyzed words per main clause, for example, also explored subordinate clauses per main clause:

Subordinate Clauses per Main Clause by Grade Level

Grade
Level
Loban's
Study
Hunt's
Study
O'Donnell's
Study
3     .18
4 .19 .29  
5 .21   .27
6 .29    
7 .28   .30
8 .50 .42  
9 .47    
10 .52    
11 .45    
12 .60 .68  
Professional
Writers
  .74  

     The large increase between seventh and eighth grade led Hunt and some of his colleagues to conclude that subordinate clauses are mastered in seventh grade. This is an extremely provocative and complicated question that I cannot discuss here in detail. It is interesting to note, however, that in my experience seventh grade teachers are the ones who are most likely to complain about the comma-splices, run-ons, and fragments in their students' writing. These are all clause-boundary errors that could be the result of their average and below-average students struggling to get subordinate clauses into their writing.
     Also interesting and relevant here is O'Donnell's concept of "formulas"--strings of words that children master as wholes without total mastery of the grammatical construction. By the time they enter school, for example, most children have used subordinate clauses as direct objects thousands of times after "formulas"" such as "Daddy said I could go." Similarly, they may learn and use many adverbial clauses as strings -- "When it gets dark, come home." My point here is that if the results of these studies are valid, they pose a serious question about what we should expect from--and what grammar we should teach--to students before they enter seventh grade.

     Unfortunately, the work of these researchers was abused as some educators began to assume that longer equals better. Thus, many of the studies that supposedly show that teaching grammar is useless (or even "harmful") were based on sentence-combining exercises and then considered the longer sentences as simply better--even if they contained more errors.
     The trend toward sentence-combining led to many teachers simply bringing sentence-combining exercises into their classrooms -- for everyone to do. The teachers were almost always unaware that in the studies that claimed sentence-combining is better, errors in the students' writing had been eliminated before the final results were tallied. In one study that I am aware of, the errors tripled in the writing of the students who did the sentence-combining."  And, as might have been expected, sentence-combining is most effective with those students who are already good at combining shorter sentences into longer ones. [1] When such exercises are brought into the classroom for everyone to do, they simply push all students into writing longer sentences, thereby, perhaps, pushing good writers into longer and weaker sentences.

     The KISS Approach, of course, enables students to see what, how, and why when they are combining sentences so that errors will not increase, but statistical exercises in KISS grammar also enable students to see where they themselves are in relation to their classmates (and everyone else for that matter). If nothing else, students can be given the results of the studies by Hunt, O'Donnell, and Loban (above). Then, instead of an emphasis on longer, longer, and longer sentences, most students should be encouraged to aim for the average. If, for example, they are between eighth and eleventh grades, they (and their teachers) should be satisfied if they are averaging ten words per main clause. Instead of pushing for more length, the instructional emphasis should be on sentence variety, and control (i.e., avoiding errors.) With that control, they will progress, naturally, into longer main clauses.
     In the KISS Approach, students can start doing such studies of their own writing as soon as they are fairly comfortable in KISS Level 3 (Clauses). In the approach, students put a vertical line at the end of each main clause. To arrive at a figure comparable to that in the studies, all they have to do is to count the words in the passage they wrote and are analyzing, and then divide that number by the number of vertical lines. Most students will find themselves pleasantly pleased. Some, however, will see for themselves that they are below the norm, and, human nature being what it is, they will probably want to catch up, especially since the KISS Approach can give them good, usable guidance for doing so.
      The students that find themselves well above the norm raise some additional questions. The first two are How much above the norm are they, and how error-free is their writing? If their writing is basically error-free, and they are not much above the norm for professional writers (20 words per main clause), then they are fine. If their writing contains numerous errors, they should be encouraged to simplify and gain control.
    My college Freshmen often did such a study. As a class, they always averaged between 14.9 and 15.5 words per main clause. But I usually had three or four students who average close to 25 words per main clause. These students are, I firmly believe, hurting themselves. The KISS psycholinguistic model helps students understand how and why. The model suggests that we process incoming information in a very tight, seven-slot, working memory. Within those seven slots, we probably handle not just the syntactic "chunking" of the sentence, but also some global questions -- such as the point of the entire paper, the topic sentences, etc. Any crash  in the processing may therefore cause a reader to lose track of important points of the paper. And the longer the main clauses are, the more likely it will be that some readers will have trouble processing them. An error that might be minor in a short main clause can cause a major crash in a 30-word main clause. Students understand this, and thus statistical exercises can put a brake on the push for more and more length. And, of course, the KISS Approach includes exercises in de-combining as well as sentence-combining.
      The National Council of Teachers of English has often claimed that students have a right to their own language, but that right is meaningless unless students have some perspective on how their language, their writing, compares with everyone else's. Statistical exercises can give students that perspective.

     Kellogg Hunt raised another very interesting point in his "Early Blooming and Late Blooming Syntactic Structures." [2] In essence, he claimed that most high school students use few, if any, appositives or gerundives. Both of these constructions can be seen as reductions of subordinate clauses.

Subordinate Clause: Martha, who is a high school senior, wrote an excellent paper on nuclear physics.
Appositive: Martha, a high school senior, wrote an excellent paper on nuclear physics.

Subordinate Clause: For a long time he struggled, as he tried to get the egg to go through the neck of the bottle.
Gerundive: For a long time he struggled, trying to get the egg to go through the neck of the bottle.

In an introduction to statistical studies, I cannot get into all the questionable aspects of this study, but my own research supports it as does developmental theory -- students cannot very well master the reduction of subordinate clauses before they master subordinate clauses themselves.
     Hunt's essay is one of the primary reasons for KISS focusing on clauses in Level 3, and leaving gerundives (and other verbals) to Level 4. Appositives are in Level 5. (The other primary reason is that almost any text will include more clauses than it will gerundives or appositives.) Another nice aspect of Hunt's idea of "late blooming" constructions is that it enables teachers to praise the "advanced" constructions that do occasionally appear in the writing of even the weakest student writers. 

     For students, the value of doing a statistical analysis of their own writing probably cannot be overstated, especially if it is done in the context of their classmates’ writing, or, if that is not possible, in the context of the research studies discussed above. One advantage is that counting constructions makes them look at the syntax of their own writing much more closely than they normally would. For example, once they learn how to identify prepositional phrases, students can place them in parentheses almost without thinking about them. Counting the prepositional phrases, however, requires more time, but also provides a different perspective—how many do they actually use? This becomes even more interesting if they can compare the number they use to what their classmates are doing. In other words, let the students analyze their own writing and then compare it to a norm.. 
     I used to have my college Freshmen analyze a sample of their own writing for words per main clause and for subordinate clauses per main clause. One class period was spent in small-group work with the students checking each others' analyses and statistics. In was not unusual for a student to bring her or his paper to me and say, "Doctor Vavra, I don't have any subordinate clauses." A quick check verified that, and I suggested sentence-combining exercises from the KISS site. The students appeared to take the problem and the suggestion seriously, especially since they could see for themselves, from what was going on in the class, that most of their peers had at least a few subordinate clauses in their samples.
     They could also see that other students were coming up to me to ask, "I have a subordinate clause within a subordinate clause that is itself within a subordinate clause. Is that o.k.?" In such cases, my answer was usually, "Yes, but you might want to consider some de-combining exercises." It was, I knew, near the end of the semester and most of these students would never have formal work on grammar again. Few of them probably used my suggestions. But the point is that these students were beginning to see and understand some basic aspects of their own writing styles in the context of the writing of their peers. Students should probably do at least one such statistical analysis of their own writing every year. And these studies should be kept so that the students can see for themselves how their writing styles change as they grow older. 
    Perhaps an even more important example is a retired gentleman who was in an advanced essay course that I was teaching many years ago. He wanted to write a book, but he said that first he needed to improve his writing. The class met once a week, and after most classes, he and I would chat about his writing. I couldn't find any problems with it, and I kept probing to see what he thought his problem was. Finally, he stated that one of his teachers had told him that his sentences were too long. As soon as he said that, I knew what to do.
     We took several samples of his writing and simply counted the number of words per main clause. We then compared the result (21 words per main clause) with those of Hunt, O'Donnell, and Loban. There was, in essence, nothing "long" about this gentleman's sentences. But a subjective comment by one of his teachers resulted in his feeling insecure about his writing not only throughout the rest of his education, but also throughout his entire professional career! Teachers, often without thinking, can do that. I have often heard teachers refer to sentences as being "short and choppy," although I myself have no idea of what they mean by "choppy." Subjective comments can hurt students, often seriously.
     Statistical research, done by  students on their own writing, and done in the context of that by Hunt, etc. (and of some on this site) eliminates the subjectivity. And, as noted above, the objective of the project is not only to enable students to see how their writing matches the "norm," but also to keep their writing within a reasonable range of that norm.



1. See "Words Enough and Time: Syntax and Error One Year After," by Elaine P. Maimon and Barbara F. Nodine.  and "Sentence Expanding: Not Can, or How, but When," by Rosemary Hake and Joseph M. Williams in Daiker, Donald A., Andrew Kerek, & Max Morenberg, eds. Sentence Combining and the Teaching of Writing: Selected Papers from the Miami University Conference, Oxford, Ohio, October 27 & 28, 1978. The Departments of English, University of Akron and the University of Central Arkansas, 1979.
2. "Early Blooming and Late Blooming Syntactic Structures." In C.R. Cooper & L. Odell (eds.) Evaluating Writing: Describing, Measuring, and Judging. Urbana: NCTE, 1977. 91-104.

See also:
The Problem with Pure Statistics: A Closer Look at an Eighth Grader's Writing --"The Road to Lhut Golane," by Kellen in Australia [Grade 8, Feb. 15]