GIST: Evolution

GIST *Helps you prepare & present research reports & proposals

A GIST Report

How GIST evolved: the need for coherent science writing

The internet has revolutionized communication in science, allowing a more rapid, widespread and interdisciplinary sharing of information than ever before. In the case of published papers, millions of research articles can now be accessed from the remotest location, and with the increasing popularity of open access journals, more and more of these articles are available for free.

However, the potential of these changes to benefit science is not being fully realized. While access to articles has improved enormously, I believe few scientists would feel there has been any concurrent improvement in their average intelligibility. As noted anecdotally by Cooper, 2008: “However, despite these advantages [of internet publication], there is a growing risk that papers published today are less successful in meeting their objectives than in the past”. There are no doubt many reasons why many if not most scientific articles are difficult to read, and I will not speculate about them all here. But Gopen and Swan (1990) argued convicingly that unintelligibility is not inevitable in science writing. Drawing on research into the expectations that readers have about the structure of an expository text, they provided some extremely useful suggestions for authors, mostly applicable at the level of the sentence and paragraph (also see: (Gopen 2004)).

During the last 12 years I have been running workshops on scientific writing, and acting as a writing guide to students and academics at the University of NSW, (Sydney) and the National Centre for Biological Sciences (Bangalore; see my online writing course, and my full CV). In this time, one of my main objectives has been to understand how writers of scientific papers could take advantage of what is known about the reader’s expectations of structure above the level of the paragraph, in particular for the paper as a whole. There are good reasons to try to do this. With respect to expository text in general, it is well accepted that as (proficient) readers work through a text they are constantly trying to fit the information into an hierarchical structure (McKoon 1977; Kintsch 1992; Pressley 2000; Dymock and Nicholson 2007). They are thus employing what is generally referred to a “top-down” approach. As noted by Spyridakis, (2000): “When readers possess a pre-existing content or structure schema that is relevant to the information they are reading, the reading task is easier and more successful.”

Scientists thus have a vested interest in the answer to the question: what “pre-existing content or structure schema” do readers bring with them to a scientific paper? To my knowledge while some research has provided information pertinent to this question, there has not been any comprehensive answer. Some might argue that readers expect a paper to conform to the much-used IMRAD system (Introduction, Methods, Results and Discussion), and while this is true, IMRAD seems to operate more like a collection of schemas, one for each of the four sections. Meadows (1996) proposed that the IMRAD format has evolved to facilitate modular, rather than linear, reading habits (also see Sollaci and Pereira (2004)). Perhaps this trend has gone too far? Many young scientists, writing up a paper, remain very uncertain about the inter-relationship of the sections, at least with respect to the sorting of content between the Introduction and Discussion. This suggests that, one step back, the papers they are reading either do not adhere to any overall hierarchical structure, or if they do, it is too indiscernible to make any lasting impression.

Should we perhaps resign ourselves to a post-modernist future for the scientific literature, in which any desire for textual coherence is viewed as outmoded and unrealistic? While I have not yet personally encountered any discussion of this issue in relation to science writing specifically, it is something that is being thought about, and debated, very seriously in relation to the communication of information generally on the world wide web. The web of course represents the ultimate expression of modular communication, and any given browsing session will almost inevitably yield information that does not sit easily within a single, hierarchical framework. Nevertheless, various researchers (e.g. Charney 1994; Spyridakis 2000) have recommended that authors of web texts should recognize that readers still expect that any individual text on the web will be conceptually coherent.

So, what I have tried to argue to this point is that (1) the content of a scientific paper should be hierarchically organized, thus satisfying a most basic requirement for overall coherence; (2) the IMRAD schema, despite all its other benefits, does not seem to encourage or ensure coherence and (3) there does not appear as yet to be any other suitable schema.

(The rest of this section is shown in point form only)

1. One section of the scientific paper (Introduction) has a well-studied known hierarchical framework (Swales, 1990)

2. The Introduction framework of Swales can be scaled up, and reflected upon itself, to provide a single framework for the paper as a whole (the GIST system)

3. The benefits of planning a paper using the GIST system:

a. It provides a backbone of coherence for the entire paper, using a dichotomy (question-answer) that is central to the process of scientific investigation.

b. One naturally intuitive way of progressing through the diagram (i.e. down and up the "V") recapitulates, for the most part, the chronological progression of the paper (and the investigation) itself.

c. On the questions side, GIST would help to ensure that the writer will think about (and explain) why they have chosen any particular sub-question (these explanations would be accessed via the "i" buttons).

d. On the answers side, GIST would help to ensure that the writer recognises the need to address why they made any particular inference. For example the writer might add a comment stating that, based on similar evidence in another species, another author has come to a similar conclusion (these additions to the argument would be accessed via the "i" buttons).

e. The GIST diagram is a useful tool to discuss the content of paper with collaborators, before writing a draft.

4. The GIST system provides a framework for further development of the paper using another, more psychological, hierarchical system (the story)

5. The GIST system can be used equally well with planning a seminar or a proposal.

References

Charney, D. (1994). The impact of hypertext on processes of reading and writing. Literacy and computers. S. J. Hilligoss and C. L. Selfe. New York, Modern Language Association: 238-263.

Cooper, L. (2008). "Better writing and more space needed online." Nature 455: 26.

Dymock, S. and T. Nicholson (2007). Teaching text structures: A key to non-fiction reading success. New York, Scholastic.

Gopen, G. (2004). The Sense of Structure: Writing From a Reader’s Perspective. New York, A.B. Longman, Pearson Education Division.

Gopen, G. D. and J. A. Swan (1990). "The Science of Scientific Writing." American Scientist 78(6): 550-558.

Kintsch, W. A. (1992). A cognitive architecture for comprehension. The study of cognition: Conceptual and methodological issues. H. L. Pick, P. van den Broek and D. C. Knill. Washington, American Psychological Association: 143-164.

McKoon, G. (1977). "Organization and information in text memory." Journal of Verbal Learning and Verbal Behavior 16: 246-260.

Meadows, A. J. (1996). Communicating research. San Diego, Academic Press.

Pressley, M. (2000). What should comprehension instruction be instruction of? Handbook of Reading Research. M. L. Kamil, P. B. Mosenthal, P. D. Pearson and R. Barr. Mahwah, NJ., Lawrence Erlbaum. 3: 545-561.

Sollaci, M. S. and M. G. Pereira (2004). The introduction, methods, results, and discussion (IMRAD) structure: a fifty-year survey. Journal of the Medical Library Association 92: 364-367.

Spyridakis, J. H. (2000). "Guidelines for Authoring Comprehensible Web Pages and Evaluating Their Success." Technical Communication 47: 310-310.

Swales, J. M. (1990). Genre Analysis: English in Academic and Research Settings Cambridge, Cambridge University Press.

GIST FAQ

What do the colours of boxes mean?

1. All the colours of the "questions" are within the red-orange-yellow range, all those for answers are within the green range. I have chosen that particular colour range contrast with the hope that it resonates with the red-green system used for stop-go, problem-solution, etc. in some other contexts.

2. Within any one range, the "intensity" of the colour is meant to indicate the level of the question or answer within the hierarchy. This is also signalled by (a) the degree of indenting and (2) the shape of the boxes. I could not have a different shaped box for each level, because of limitations of the mapping program I am using (MindMap, otherwise a very fine, if fiddly, product!). The use of potentially-confusing colors may be made unnecessary if good graphic design can make the hierarchical level of an element unambiguous.

Why do the answers exactly mirror the questions?

The nested Q&A mirroring is at the heart of the system, and is meant to help ensure that researchers:

1. Come up with a tightly integrated unit of communication

2. Do not pose questions that they don't address at some later stage

3. Do not, for the most part, provide answers for previously unforeshadowed questions. While I think it's fine that some implications are not foreshadowed, the main focus of the Discussion should be to answer previously introduced questions. It's a typical error to get carried away in the Discussion with highly speculative issues.

The questions always lead to answers, but can't answers can also lead to questions?

I agree, but in both of the cases I consider below, this does not lead to any need to allow this within GIST. There may be other cases where it does!

Case 1: In some papers, an hypothesis is proposed in the Intro, and then rejected in the early part of the Results, on the basis of contradictory findings. Then, still within the Results, an alternative (and previously un-introduced, and probably unimagined) hypothesis becomes the focus of some subsequent experiment, reported upon in a later part of the Results. Thus should the template have a provision for a Q1 > A > Q2 type of progression? I think the answer is no because I see that the template's main job is to capture the logical, rather than the chronological, flow of the project. In which regard, both Q1 and Q2 are *logical* daughters of Q1's already existing higher-order question, even if, for Q2, this only became apparent to the researcher after Q1 had been answered. (Of course the chronological order may need to be made clear in the text of the paper itself, particularly for example, if both (1) the same data set is being used to address the two questions and (2) the optimal experimental design for Q2 is different from that used for Q1.)

Case 2: In the Discussion, answers from the current study can of course suggest questions to be addressed in the future. But I would not see such questions as central enough to be included in the GIST diagram. In support of this view, I think that questions of this type would rarely find their way into a paper's Abstract.

In the example of paper diagrammed using GIST -on the Picobar Tree Frog- lower-level answers only ever connect to answers one level higher. In general, are there not more possibilties?

It is true that, in principle, answers could "talk to each other" in more ways than shown in my diagrams, and I think that flexibility in this aspect is required in the program. The most obvious counter-example I can think of comes from mathematical proofs, where very un-nested flows of reasoning can occur. Nevertheless, I think that in scientific research, the highly nested flow pattern shown in my diagrams is far and away the most typical case, for both questions and answers. In science, arguments (as opposed to explanatory models) are usually very basic, and if they seem complicated it's probably because of the author's poor communication skills! And therefore I think there is much value in making the flow pattern I have used here the default, but not required, pattern.

If you have any Questions or Comments please leave a comment in the guestbook, using the link below, or email Geoff Hyde at the National Centre For Biological Science, Bangalore: geoffATncbsDOTresDOTin