- A scheme for writing a research paper in collaboration, or by oneself
This is from Prof Volker Heine from the TCM group at the Cavendish Laboratory in Cambridge.
A scheme for writing a research paper in collaboration, or by oneself
Fri, 19 Feb 2016
The following is the scheme I have arrived at through trial and error, to be most efficent for collaboration on a research paper, e.g. with one's supervisor. It is a scheme for organising one's thoughts so that I also use it when writing by myself. Incidentally, for Ph.D. students, it is best to write the research papers *before* the thesis, because the papers can be written in collaboration with the supervisor, but (s)he is not supposed to write the thesis!
THE PRINCIPLE IS TO *first* AGREE VERY VERY CLEARLY WHAT TO SAY, *before* PRODUCING ANY TEXT. THAT IS THE HARD PART, but it is very frustrating for both sides to receive a draft manuscript, which has taken a lot of time to produce, and then say "Well yes, but the main ideas do not come through to me very clearly", or "It doesn't quite hit the nail on the head."
The scheme goes through three stages, and THE OBJECTIVE AT STAGES I AND II IS TO PRODUCE *a list of points* IN THE RIGHT ORDER, EACH STATED *very precisely* IN ONE *complete sentence* WITH OR WITHOUT A SKETCH FIGURE. Each stage must be agreed *before* going on to the next stage. It is only by writing a complete sentence that one properly expresses a thought: just jotting down a heading only hints at it. The effort in writing these sentences is not wasted because most get incorporated in the next stage.
The final stage III is then EASY: these sentences just get expanded into paragraphs and sections.
See also the four more detailed Appendices on Figures and style etc..
The problem in all writing, lecturing, making a presentation to a committee or persuading one's boss, is to make the main points stand out clearly and simply from all the detail. That is the constant objective in the scheme.
Stage I. The bare outlines
I.1. Write the title. Almost all readers scanning the journal contents will not read any further than the title, unless it engages their particular interests. So the title has to be clear and meaningful.
I.2. Write the abstract (but to be revisited at the end and revised then if necessary). (a) The purpose of the paper. (b) Include key words that might be used in a literature search.
I.3. Read Appendix C on Introduction and Conclusions.
I.4. Now list the section headings for the paper.
I.5. List the main points to be made in the paper, itemised, in about one page total.
I.6. Authors, in what order?
I.7. What journal will it be submitted to?
Stage II. Structuring the paper
After stage I is complete and agreed by all authors.
II.1. Read Appendix A on Figures.
II.2. Read Appendix C on Introduction and Conclusion.
II.3. EXPAND EACH SECTION FROM I.4 INTO A SERIES OF POINTS, IN THE RIGHT ORDER, WITH *one complete clear sentence for each point*, the whole occupying up to one page. At this stage it should be readable as a coherent story by someone highly knowledgeable in the field. See the examples given in Appendix B.
II.4. Include in that page LITTLE SKETCHES OF ALL FIGURES WITH CAPTIONS. PLAN THE FIGURES TO TELL THE STORY -- see Appendix A on Figures. Remember that most readers read the abstract, perhaps look at the introduction, and then 'read' the figures as the quickest route to the results.
II.5. Create a detailed 'editorial style kit' as per Appendix D for the journal to which the paper will be submitted.
II.6. How will the paper be submitted? In Tec? Does the journal have a Tec template for one to use?
Stage III. Now (and only now) write the paper
III.1. Read Appendix D on the journal's required style for references, figures etc.. Use it to get these editorial bits correct in the first shot. It is a pain to tidy them up later, and one overlooks points.
III.2. Read Appendix B on style of writing. Reflect and take deeply to heart that at *every* MOMENT THE MIND NEEDS TO KNOW IN WHAT DIRECTION YOU ARE WANTING TO LEAD IT.
III.3. Read again Appendix A on Figures, and reflect on how one can design a figure to tell a story. Reflect on how a good advertisement has the bare minimum of information, only that which you must know.
III.4. Take each point from II.3 and expand it into a paragraph. Here the single sentence you created in II.3 above will often become to opening 'pointing' sentence of the paragraph.
III.5. At the same time as writing the paragraph, design any figure and caption that are part of it. In fact where there is a figure, it is best to do the figure as more important, before writing the paragraph.
III.6. Finally look again at the proposed title and abstract from items I.1 and I.2, and revise if necessary.
III.7. Send it off to the journal! If stages I and II were carried out well, then there should not be any need for much discussion with co-authors at this point.
A. Figures B. Style of writing C. About Introduction and Conclusions D. Editorial style kit
Appendix A. Figures
Think like an advertising executive: a good picture conveys information better than a million words.
Each figure has to be carefully designed. What is the *main* point being made? Is it the calculated result, or the comparison between calculation and experiment? Are the absolute values of the quantities and scales important, or are relative values sufficient? Error bars or not? In a diagram, for example of a crystal structure, how much *needs* to be shown: probably not every bond to every atom! What aspect of the structure is important for *your* story?
The unnecessary or unimportant aspects can often be pruned right out of a diagram, or shown in a simplified stylised form.
Labelling and caption should be self-explanatory *without* needing to read a whole load of text. Here using *conventional* or customery symbols for quantities is a great help. Some authors blatently disregard that: they use any damn symbol and think that is OK if they defined it 10 pages back in the middle of a paragraph!
Look at an issue of 'Scientific American', which takes this matter of designing figures and captions to an extreme. (They don't just print what an author has sent in!) They design figures and captions to ensure the *whole* story can be obtained and understood by *only* reading the figures. See how it is done! You will not be allowed by the journal to make your captions quite as long as in Scientific American, but you can learn from their style.
Appendix B. Style of Writing
The human mind does not cope easily with *disconnected* information. Some people, often in the mathematical and physics and computing worlds, have a particular problem in this direction, as I do: that is why we have gravitated into the most one-track minded subjects in the world! But in a more general way it is universal.
This applies at all levels, from the overall to the detailed. For example the mind wants to know what the question is before receiving the answer. *Contrary* to mathematical logic, it likes to know and understand the result *before* all the tortuous path leading to it.
I tend to liken it to a coat hook, which has to be there and in place, ready to have subsequent information hung onto it. I believe it is called the style of Macauly (19th century writer and politician) which has been used in training the British Civil Service, and which I learnt at school.
Another aspect of the style is to use a variety of sentence lengths, none very long. A very short one can emphasize a point without using bombastic language. It arrests the attention. (Note how I have used just 4 words.)
Just follow this drill. (Again I have used just 4 words, re the purpose of the next few sentences.)
(a) The introduction section to the *paper* has to say clearly why the research is being done, what the issues are, the main ideas, what one hopes to achieve, and how/why. See Appendix C on introduction and conclusions.
(b) The first paragraph of each *section* says what the section is about, simply.
(c) The opening sentence of each *paragraph* says what the paragraph is about. Very often one can use the relevant sentence from step II.3.
Here are some examples of sentences opening paragraphs. (i) The energy expression (6) in our case has to contain the extra (and unusual) term mu. It takes into account .... We derive it .... (ii) We turn now to consider what happens when the temperature varies. (iii) In order to calculate a phase diagram, we need the energy DeltaEm expressed relative to the energy of a mechanical mixture of SiO2 and Al2O3. The latter ....
There is sometimes an extra wrinkle about this which one needs to watch (note: outlines the purpose of this paragraph). When developing some argument, the conclusion, i.e. the main point, naturally comes at the end of the development, which would place it at the end of the paragraph. Unfortunately, people's attention tends to focus on the beginning of a paragraph, so that the all-important conclusion at the end tends to get missed! In fact, one normally wants it to feed into the topic of the next paragraph, and I use two slightly different techniques to overcome the problem. One cheeky way is simply to take the last sentence of the development, remove it from where it naturally would come, and simply put it as the first sentence of the new paragraph, with maybe an extra few words such as 'Thus we conclude that ...'. The other way of dealing with the problem is to have the conclusion at its logical place at the end of the development, and then simply to repeat the conclusion at the start of the next paragraph using slightly different words to disguise the repetition, maybe with a few extra words to help, e.g. 'We have seen that ...'
Appendix C. About Introduction and Conclusions
The main purpose of the Introduction must be to plug the readers clearly into seeing what the issues are. They cannot appreciate the answers unless they are clear about the questions. Hence I have sometimes titled it Introductory Survey or even Survey of the Issues and Conclusions.
Occasionally once having thoroughly explained the issues, the conclusion has simply been 'Yes' (or 'No'), or 'A' (rather than 'B'), in which case it seemed absurd to have a separate Conclusions section, and I have just splilt the beans in the Introduction.
In any case this seems much more important than the boring history of who has previously done what, which the reader will know if they are sufficiently into the research field to read the paper. (Sorry about the grammar, but it seems that now-a-days 'they' is used for the singular 'she or he'.)
However in some fields of research it is customery to include quite a history, and some journals and referees insist on it (especially if their work has not been mentioned).
Appendix D. Editorial style kit
This is boring but necessary, and as remarked in Sect. III.1 it is best to get the details right first shot, rather than through revision later.
It is equally essential to be familiar with how the paper is to be presented electronically for publication (see Item II.6), including the figures.
Each journal has its own 'house' style for figures and references etc. which editors feel very strongly about, and there is no point in irritating them.
Therefore make yourself a style-kit, listing the required form for the following. This may already exist on the journal's website.
Section headings: enumerated by Roman or Arabic numbers or letters A, B.., followed or not followed by a full-stop, with capital letter for each word or not.
Ditto table headings. Tables numbered 1, 2, 3, .. or I, II, III etc..
How are figures and tables referred to in the text, both when used as a noun in a sentence and when in parenthesis. E.g. ... where Figure 2 shows how ... (Fig. 3). Figure or figure, Fig. or fig., Table or table.
Similarly Equation, equation, Eq. or eq..
References. This is a complex subject if references are not numbered 1, 2, 3 etc.. If they use the style 'Smith (1998)', then with two or more authors in some journals all are given the first time and then subsequently just Smith et al.. When there is Joe, Blogs and Smith (1990) and also Joe, Blogs and Jones (1990), the first may be Joe et al. (1990a) and the latter Joe et al. (1990b), but there are variations.
Does the journal have some idiosyncratic spellings such as 'connexion'?