For almost two decades I have been telling students that one of the reasons for studying grammar is to improve one's control over the meaning of what one writes. But what, exactly, is "meaning," and more importantly, from the point of view of research, how can one measure it? I had fumbled with these questions for many years before Rochelle McAndrews, now a graduate of our Technical Communications program, and also one of the brightest students ever to grace my classroom, introduced me to the concept of "idea units" in articles by Carol S. Isakson and Jan H. Spyridakis. Although the definition of "idea units" used by Isakson and Spyridakis is too broad for my purposes, the fundamental idea provides a method for measuring meaning. But before the measurement could in any way be meaningful in assessing students' writing, I needed a corpus of ideas that were to be meant. The Aluminum passage provides that corpus.
     Just as nails can be pounded into any piece of wood, any text can be broken into idea units. And just as nails sticking out of pieces of wood can be dangerous, so can texts broken into idea units. Without a purpose, such broken texts simply add to the academic confusion. Isakson and Spyridakis were interested in assessing how various syntactic structures, among other things, affect readers' comprehension and retention of information. They therefore broke published articles into idea units, had subjects read the articles, and tested them on their comprehension by counting the idea units that the readers retained. But whereas they were interested in the decoding of texts, my interests in in how writers (particularly students) encode texts -- how do they put ideas onto paper? In order to measure this, it helps  to know the ideas in advance, and the Aluminum passage provides just such knowledge. The questions now become can we determine differences in the ideas which become encoded and in the logical relationships among those ideas?

The Definition of Idea Units

     In the study reported in "The Influence of Semantics and Syntax on What Readers Remember," Isakson and Spyridakis used two passages from  Scientific American of approximately 1,330 words each. They explain how they determined idea units:

So that we could measure recalled ideas against ideas in the texts, we quantified the information in each text by separating the texts into idea units (IUs), using a method initially developed by Johnson that we have refined (Spyridakis and Isakson 1998;  Wenger and Spyridakis 1993; Spyridakis 1989). IUs are considered to be the smallest information units that readers may logically pause at to emphasize, to enhance meaning, or to take a breath; they are not based on text structure (Johnson 1972). To separate the texts into IUs, 26 students (13 per text) in an advanced technical communication class parsed (separated) all ideas that could be construed as IUs. The students were given the definition of IUs and a sample text with its parsed IUs to help them understand IU parsing. IUs were deemed to exist when a majority of students agreed on the parsing location. As an example of IUs in our study, note the following sentence, excerpted from the Sensors text, and its IU distribution.

Each of the three main arthropod groups -- insects, arachnids and crustaceans -- has its own version of a sense organ for detecting deformations or strains in its cuticle, or exoskeletal material.

IUs
1. Each of the three main arthropod groups
2. -- insects, arachnids and crustaceans --
3. has its own version of a sense organ
4. for detecting deformations or strains in its cuticle,
5. or exoskeletal material.

      In addition to simply counting recalled IUs, Isakson and Spyridakis attempted to distinguish more important ideas:

Next we determined the importance of the IUs by using a majority tally from the 26 students who, two weeks later, rated the importance of each IU in relation to the overall meaning of the passage. Reading the passage they had not parsed, they labeled more important IUs with the number 1 and less important IUs with the number 2.

Twenty-two logical connectives (conjunctions, conjunctive adverbs, and conjunctive phrases), chosen from Halliday and Hasan's list of coherence devices [6], were added to each text. Words such as "however," "generally," "in fact," "also," "next," etc.,. were placed among idea units at all levels. {Spyridakis, 398)

Logical and Rhetorical Relationships between Idea Units
 
 

"Missed" Idea Units

The following table indicates how many of the 93 writers omitted each of the indicated idea units.
 

Omissions at the End of the Passage

     The most frequently omitted units (nineteen times) are numbers 29 ("The metal is light.") and 30 ("It has a luster."). In part, these omissions result from students either not finishing or rushing to finish. (In addition to specific revisions that were not completed, note the high number of omissions beginning at unit 28 and continuing to the end.) There is, however, at least one other probable reason for some of these omissions -- beginning with unit 29, the passage may be confusing and/or redundant.
     The confusion starts with "The metal is light." Does this mean light in color, or in weight? Two of the students (#23 and #50) felt the need to clarify this by adding "in weight." "Light": is then followed by "luster," "bright," and "silvery." It is probable that some students felt that "luster" implies "bright," or that "bright" implies "luster," or that "silvery" implies both "luster" and "bright." It may have been this felt redundancy, rather than a rush to finish, that accounts for the omission of one or more of these units.
     The term "luster," moreover, clearly gave some students problems:

# 3 "This metal is light, luster, and comes in many forms. The luster of this metal is both bright and silvery."

# 26 "The result is a bright silvery luster that comes in many forms."

# 64 "After this process the final product is a metal that is light and luster and that looks silvery and bright."

# 69 "Finally, a metal with light and luster that is bright and silvery is formed."

# 72 "The workmen use electricity to produce a metal that is light, bright, silvery and luster."

General/Specific Omissions

     Idea unit # 10 ("Workmen extract these other substances from the bauxite.") was omitted by eleven students, but its omission removes no "ideas" from the passage. In effect, unit 10 is a generalized, introductory statement to the specific ideas presented in units 11 through 19. It is probable that some students felt that this generalization was unnecessary. If this was, in fact, the case, it raises an interesting question about readability.
     I do not claim to be an expert on readability formulas, but from what I have seen, they have two legs -- a thick long one (vocabulary), and a short thin one (grammar/sentence structure). The omission of unit 10, however, points to a more global question of readability. For some readers, unit 10 may be superfluous, but other readers may find this part of the passage easier to understand if it is preceded by the generalization in unit 10. The generalization may not only provide these readers with a perspective on what is coming, but it may also simply slow down the rate of incoming specific information, thereby giving these readers time to digest it.
     The question is complex, and in part it depends on the readers' previous familiarity with the topic of the passage, but the question surely deserves further study. Can a readability formula be considered valid if it ignores this variable? 
     If unit 10 was omitted because it is a generalization of units 11 through 19, we also need to note the possibility that some of the omissions of units 11 through 19 may be the result of the writers feeling that they were covered by the inclusion of unit 10.

Isakson, Carol S., and Jan H. Spyridakis. "The influence of semantics and syntax on what readers remember." Technical Communication. Aug. 99 Vol. 46, No. 3, 366-381. Downloaded from ProQuest  Direct 2/2/00.

Spyridakis, Jan. H. "Signaling Effects: Increased Content Retention and New Answers -- Part II. Journal of Technical Writing and Communication, Vol. 19 (4), 395-415, 1989.