Frequency of Formal Errors in Current College Writing, or Ma and Pa Kettle Do Research

 

 

404                    College Composition and Communication 39 (December 1988)

marked only once for every four times they appear. The number of errors found compared to the number of errors marked suggests a fascinating possibility for future research: detailed observation of teacher marking, accompanied by talk-aloud protocols. Such research seems to us a natural follow-up to the findings presented here.9

fThird, the reasons teachers mark any given error seem to result from a com-piexiformula that takes into account at least two factors: how serious or annoying the error is perceived to be at a given time for both teacher and student, and how difficult it is to mark or explain-As Table 1 shows, the errors marked by the original teachers on our papers produce a different (although not completely dissimilar) ranking of errors than the formal count we asked our raters to do. Some of the lesser-marked errors we studied are clearly felt to be more stylistic than substantive. Certain of fhe comma errors seem simply not to bother teachers very much. Others, like wrong words or missing inflections, are much more frequently marked, and might be said to have a high "response quotient" for teachers. In addition, we sensed that in many cases errors went unmarked not because the teacher failed to see them, but because they were not germane to the lessons at hand. A teacher working very hard to help a student master subject-verb agreement with third-person singular nouns, for instance, might well ignore most other errors in a given paper.

Teachers' perceptions of the seriousness of a given error pattern seem, however, to be only part of the reason for marking an error. The sheer difficulty of explanation presented by some error patterns is another factor. Jotting "WW in the margin to tip a student off to a diction problem is one thing; explaining a subtle shift in point of view in that same marginal space is quite another. Sentence fragments, comma splices, and wrong tenses, to name three classic "serious" errors, are all marked less often than possessive apostrophes. This is, we think, not due to teachers' perception that apostrophe errors are worse than sentence-boundary or tense problems, but to their quickness and ease of indication. The its/it's error and the possessive apostrophe, the two highest-marked patterns, are also two of the easiest errors to mark. This is, of course, not laziness; many composition teachers are so chronically overworked that we should not wonder that the errors most marked are those most quickly indicated.

Fourth, error patterns in Student writing are shifting in certain ways, at least partially as a result of changing media trends within the culture. Conclusions must be especially tentative here, because the time-bound nature of studies of error makes comparisons difficult and definitions of errors counted in earlier research are hard to correlate. Our research turned up several earlier lists of serious errors in freshman composition, however, whose order is rather different from the order we discovered.

Roy Ivan Johnson, writing in 1917, reported on 198 papers written by 66 freshmen, and his list of the top ten error patterns in his study is as follows (wherever possible, we have translated his terms into ours):

 

Frequency of Formal Errors in Current College Writing                           405

1. Spelling

2. Capitalization

3. Punctuation (mostly comma errors)

4. Careless omission or repetition

5. Apostrophe errors

6. Pronoun agreement

7. Verb tense errors and agreement

8. Ungrammatical sentence structure (fragments and run-ons)

9. Mistakes in the use of adjectives and adverbs

10. Mistakes in the use of prepositions and conjunctions

In 1930, Paul Witty and Roberta Green analyzed 170 papers written in a timed situation by freshmen. Here is their top ten list, translated into our terms where possible:

1. Faulty connectives                                                     \

2. Vague pronoun reference

3. Use of "would" for simple past tense forms

4. Confusion of forms from similarity of sound or meaning

5. Misplaced modifiers

6. Pronoun agreement

7. Fragments

8. Unclassified errors

9. Dangling modifiers

10. Wrong tense

As we mentioned earlier, the largest-scale analysis of errors was done by John C. Hodges in the late 1930s. Unfortunately, we know very little about Hodges' research. He never published any results in contemporary journals, and thus it is difficult to know his methods or even very much about his findings, because we can see them only as they are reflected in the Harbrace Handbook, which today still uses the exact arrangement that Hodges gave it in its first edition in 1941. In the "To the Instructor" preface of his first edition, Hodges says that his 20,000 themes "have been tabulated according to the corrections marked by sixteen instructors," which suggests that his raters looked only for teacher-marked errors (Hodges iii). In a footnote on the same page, Hodges gives the only version of his top-ten list ever published:

1. Comma

2. Spelling

3. Exactness

4. Agreement

5. Superfluous commas

6. Reference of pronouns

7. Apostrophe

8. Omission of words

9. Wordiness 10. Good use

That is all we know of Hodges' findings, but it does not seem unreasonable to assume that he reports them in order of frequency.

 

406                    College Composition and Communication 39 (December 1988)

In terms of how patterns of error have changed, our findings are, of course, extremely tentative. Assuming that Hodges' Harbrace list constitutes some version of the error patterns he found in 1939, however, we note some distinct changes. In general, our list shows a proliferation of error patterns that seem to suggest declining familiarity with the visual look of a written page. Most strikingly, spelling errors have gone from second on the list to first by a factor of three. Spelling is the most obvious example of this lack of visual memory of printed pages seen, but the growth of other error patterns supports it as well.10

Some of the error patterns that seem to suggest this visual-memory problem were not found or listed in earlier studies but have come to light in ours. The many wrong word errors, the missing inflected endings, the wrong prepositions, even the its/it's errors--all suggest that students today may be less familiar with the visible aspects of written forms. These findings confirm the contrastive analysis between 2,000 papers from the 1950s and 2,000 papers from the 1970s that was carried out by Gary SIoan in 1979. Sloan determined that many elements of formal writing convention broke down severely between the fifties and seventies, including spelling, homophones, sentence structure elements, inflected endings, and others (157-59). Sloan notes that the effects of an oral--and we would stress, an electronic--culture on literacy skills are subversive. Students who do not read the "texts" of our culture will continue to come to school without the tacit visual knowledge of written conventions that "text-wise" writers carry with them effortlessly. Such changes in literate behavior have and will continue to affect us in multiple ways, including the ways we perceive, categorize, and judge "errors."

Finally, we feel we can report some good news. One very telling fact emerging from our research is our realization that college students are not making more formal errors in writing than they used to. The numbers of errors made by students in earlier Studies and the numbers we found in the 1980s agree remarkably. Our findings chart out as follows:'l

Average Paper    Errors per    Errors per Study                     Year       Length         Paper     100 words Johnson                  1917     162 words       3.42        2.11 Witty & Green           1930     231 words       5.18        2.24 Ma & Pa                 1986     422 words       9.52        2.26

The consistency of these numbers seems to us extraordinary. It suggests that although the length of the average paper demanded in freshman composition has been steadily rising, the formal skills of students have not declined precipitously.

In the light of the "Johnny Can't Write" furor of the 1970s and the sometimes hysterical claims of educational decline oft heard today, these results are striking--and heartening. They suggest that in some ways we are doing a better job than we might have known. The number of errors has not gone down,

 

Frequency of Formal Errors in Current College Writing                            407

but neither has it risen in the past five decades. In spite of open admissions, in spite of radical shifts in the demographics of college students, in spite of the huge escalation in the population percentage as well asJn sheer numbers of people attending American colleges, freshmen are still committing approximately the same number of formal errors per 100 words they were before World War One. In this case, not losing means that we are winning.

Epilogos

Our foray into the highways of research and the byways of the Pentecostal Youth are over for a time, and we are back on the farm. From our vantage point here on the porch, we can see that this labor has raised more questions than it has answered. Where, for instance, do our specific notions of error come from? Can we identify more precisely the relationship among error patterns in written student discourse and other forms of discourse, especially the mass media? Could we identify regional or other variations in error patterns? How might certain error patterns correlate with other patterns--say age, gender, habits of reading, etc.? How might they correlate with measures of writing apprehension, or the "ethos," the ideology of a specific curriculum? Most provocatively, could we derive a contemporary theory of error which would account for the written behaviors of all our students as well as the marking behavior of teachers? These are a few of the problems we'd like to fret over if and when we decide to take to the research road again.

Notes

1. As an example of shifting perceptions of student error patterns, it is worth noting that Charles T. Copeland and Henry M. Rideout, writing in 1901, identified the most serious and common grammatical error in Harvard freshman papers as a confusion of the rules for use of "shall" and "will" to express futurity (7 In).

2. For a list of most of these studies, see Harap 444-46.

3. We wish here to express our gratitude to the College Division of St. Martin's Press, which graciously offered respondents a choice from the St. Martin's trade book list in exchange for 30 or more teacher-marked student papers or xeroxes of student papers. We are especially grateful to Nancy Perry, Marilyn Moller, and Susan Manning, without whose help this research could never have been accomplished. From assistance with mailings to the considerable tasks of paper stacking, stamping, sorting, and filing, they made the task possible. Their support, both institutional and personal, is deeply appreciated.

The demographics of the papers we were sent were interesting, as we found when examining them for our stratified sample. After pulling all the papers that were illegible, or were not undergraduate papers, were too short to be useful, or were clearly papers from ESL courses, we were left with 19,615 papers. We divided up the U.S. into seven fairly standard geographical regions:

(1) Northeast

(2) Southeast

(3) Midwest

(4) Mid-South

 

408                    College Composition and Communication 39 (December 1988)

(5) Plains States

(6) Southwest (including Hawaii)

(7) Northwest (including Alaska)

Here are the raw numbers of how the papers were distributed as they came in to us:

Region         1234567   Total Total number of papers 3,652 3,478 3,099 4,974 1,229 2,292 891 19,615 Total number of         61    51    54    55    18   47   14    300

teachers Total number of         47    35    40    39    14    24    7    206

4-year schools Total number of         14    16    14    16     4    23    7     94

2-year schools Total number of         44    49    48    48    18   44   13    264

state schools Total number of         17     2     6     7     0     3    1     36

private schools

Number of schools        2201111      8 with total enrollment under 1,000

Enrollment 1-3,000       9    13     7    11     3     54     52 Enrollment 3-5,000      13     5     5    14     2    7    2     48 Enrollment 5-10,000     19     9    16    10    6    74     71 Enrollment 10-20,000    14     9    13    13     1    15    2     67 Enrollment              4    13    13     6    5    12    1     54

over 20,000

4. We wanted to find out whether the sample of papers we had received mirrored the demographic realities of American higher education. If it did not, we would have to adjust it to represent the student and teacher populations that were really out there.

When we looked at The Digest of Education Statistics, we found that some of our numbers approximated educational statistics closely enough not to need adjustment. The breakdown between 4-year colleges and 2-year colleges, for instance, is 71%/29% in the statistical tables and 69%/31% in our sample. The state schools/private schools ratio is statistically 79%/21%, while our sample ratio was 88%/12%, but the over-representation of state schools did not seem serious enough to worry about for our purposes. In terms of enrollment, we found middle-sized schools slightly over-represented and very small and very large schools slightly under-represented, but in no case was the deviation more than 7% either way:

Number of schools with total

enrollment under 1,000 Enrollment 1-3,000 Enrollment 3-5,000 Enrollment 5-10,000 Enrollment 10-20,000 Enrollment over 20,000

% of students % in

nationally sample

4 2

11 13 21 25 25

17 16 24 22 18

 

We found the most serious discrepancies in the regional stratification, with some regions over-and others under-represented.

Region                       12345     67 % of students nationally             23    12   23    15    4    19  4 % of students in sample              19    18    15    25    6    12   5

On the basis of the regional discrepancy we found, we decided to stratify the sample papers regionally but not in any other way.

For help with the methodological problems we faced, and for advice on establishing a ran-

 

Frequency of Formal Errors in Current College Writing                           409

dom stratified sample of 3,000 papers, many thanks to Charles Cooper. When the going gets tough, the cough go ask Charles for advice.

5. These two examples of old-time error patterns are cited in Pressey and in Johnson.

6. The term "comma fault" was by far the most popular term to describe this error pattern until the ubiquitous Harbract seeded the clouds with its terms in 1941, advancing "comma splice," previously a term of tertiary choice, into a primary position by I960. See Lunsford, Glenn, and Connors, "Changing Pedagogical Nomenclature," forthcoming when we can all stop panting.

7. This paper, five lovingly-written pages of classic Victorian pornography, was extremely popular with the raters. Example passage: "Tammy's own arousal came with suddenness. Bill's urgent caresses kindled a delicious warmth in her flesh and then a melting trembling heat." We would quote more, but we're prudes, and this is a family magazine. For an original xerox copy of this extremely interesting piece of pedagogical history, send $25.00 and a plain brown self-addressed envelope to the Ma and Pa Kettle Go To Waikiki Fund, c/o this magazine.

The teacher's comment on this paper, incidently, was curt. "This is narration," wrote the teacher, "Sorry you didn't use analysis to explain. Remember the definition of explanatory prose?" Another kick in the teeth for Art.

8. In addition to the error-rating sheets, on which the raters kept track of errors found and errors marked, we asked them to write down on a separate list every misspelled word in every paper they saw. This spelling research is only partially tabulated and will be presented in another study.

9. We were also intrigued to find that of the 3,000 papers examined, only 276 had been marked using the letter-number system of any handbook. Handbooks may be widely used, but fewer than 10% of out papers relied on their systems. The rest had been marked using the common symbols and interlinear notes.

10. With our spelling research partially tabulated at this point, we are struck by the prevalence of homophone errors in the list of the most commonly misspelled words. The growth of too/to and theirltherd they're error patterns strongly suggests the sort of problem with visual familiarity suggested by our list of non-spelling errors.

11. These comparisons are not absolutely exact, of course. Johnson counted spelling errors, while Witty and Green and we did not. The numbers in the chart for Johnson's research were derived by subtracting all spelling errors from his final error total.

Works Cited

Copeland, Charles T., and Henry M. Rideout. Freshman English and Theme-Correcting at Harvard

College. Boston: Silver, Burden, 1901.

Elbow, Peter. Unpublished document. English Coalition Conference. July 1987. Harap, Henry. "The Most Common Grammatical Errors." English Journal 19 (June 1930);

440-46.

Hodges, John C. Harbrace Handbook of English. New York: Harcourt, Brace & Co., 1941. Johnson, Roy Ivan. "The Persistency of Error in English Composition." School Review 25 (Oct.

1917); 555-80. Pressey, S. L. "A Statistical Study of Children's Errors in Sentence-Structures." English Journal

14 (Sept. 1925): 528-35.

Shaughnessy, Mina P. Errors and Expectations. New York: Oxford UP, 1977. Sloan, Gary. "The Subversive Effects of an Oral Culture on Student Writing." College Composition and Communication 30 (May 1979): 156-60.

Snyder, Thomas D. Digest a/Education Statistics 1987. Washington: Center for Education Statistics, 1987.

Williams, Joseph. "The Phenomenology of Error." College Composition and Communication 32 (May 1981): 152-68.

Witty, Paul A., and Roberta La Brant Green. "Composition Errors of College Students." English Journal 19 (May 1930): 388-93.