Basics of scientific writing

This was a 1.5 hour-long seminar on scientific writing for research papers, at MIT, on October 7th, 2015.

• Take whatever opportunity yo can get to write, especially if you can get feedback and revise your writing.
  • It's really the revising that is the most instructive.

General comments about scientific writing

• Scientific writing: Precise, concise, object report of research
• Old habits die hard
  • start at 2am, finish by 9am deadline
  • use scholarly / serious / smart language
  • meet the page count
• New habits
  • write a first draft early
  • get feedback
  • revise it more than once
  • Use easy to understand language (suits your content)
  • Write to communicate!
• How to get started on writing?
  • Model your writing after someone in your field who is especially good
  • Read texts on scientific writing
    • "The craft of scientific writing", by Michael Alley
    • "How to write and publish a scientific paper", by Robert A. Day
    • "Technical and scientific writing", by Leslie C. Prelman
      • Not exactly the best book on sci. writ. but is searchable
• Aim your writing to form a story; a coherent, flowing, logical story
  • This tends to stick in people's heads
• Plan. Give your writing thought and time
• Make outlines.
  • Every good artist will develop a draft for his work
  • Try out various ways of laying out your paper
  • Forces you to establish the scope of your document
  • Forces you to develop a point of view.
  • Get feedback on your outline. Provide headins/table of contents and topic sentences for your paper
    • Topic sentences are very important in western writing.
    • If you get the topic sentences from papers in journals like Science, you end up with the outline of the paper.
• Revise, revise; at any point.
• Edit/Polish
  • Don't be afraid to edit severely.
  • Average # of drafts for scientific papers is 23!
  • Making sure paragraphs have one simple idea.
• Get feedback on your writing and provide feedback on their writing.
  • Constructiveness is important, otherwise people don't listen to the advice.
  • Ability to accept feedback is important. You are not a victim of injustice.
• Three aspects of writing style
  • Structure: organization, emphasis, transition, depth
  • Language: precision, clarity, forthrightness (go around the point or directly to it), fluidity, conciseness (the use of the fewest possible words to do your writing)

Structure

Prose is architecture, not interior decoration, and the Baroque is over. -- Ernest Hemingway

• The structure is what holds it all together.
• Don't get too fancy or too flowery
• Try to make it as simple and as direct as possible
• Organization is reflected in headings, subheadings and topic sentences
  • Think about how you read papers (in what order)
  • Think about how your reader might want to read your paper
  • What's the right granularity?
• Sometimes you need subheadings.
• The paragraph should be the unit of composition: one paragraph for each major idea.
  • This helps the reader: they know new paragraph means new topic.
  • Do not write single sentences as paragraphs.
• Transitions link ideas.
• Common flaw is poor logical flow, bad transitions.
• Transition are often at the beginning of the sentence
• Sometimes transitions are used at the beginning of the paragraph
• Type of transitions:
  • consequences; cause and effect
  • contrast
  • sequence and connectivity
  • similarity and comparisons
  • explanations
• Write with the reader in mind.
  • readers don't just read; they interpret.
  • your problem for not giving reader information where they expected
  • Recommended reading: Gopen and Swan, "Science of scientific writing"
• If you are reading a sentence aloud and you have to stop and take a breath, then it's too long.
• If the subject is separated from the verb by too many words, it's bad.
• The reader tends to wait for the verb to come and will go back to the subject, if you had a huge gap between them.
• Standard word positioning is: Subj + Verb + Obj

Language

• Scientific language avoids needless complexity
• Firstly, secondly and thirdly -> First, second and third
• Drop needless words
  • Already existing
  • Basic fundamentals
  • Completely eliminate
  • Empty space
  • Had done previously
  • Introduced
  • Mix together (can't mix apart)
  • Never before
  • In order to (just use 'to', but occasionally the rhythm of the sentence needs 'in order to')
• Weak vs. strong verbs
  • Made the arrangement for vs arranged
  • Made the decision vs decided
  • Made the measurement vs measured
  • (Known as nominalizing the verb => verb becomes very weak)
    • Really, stay away from "made"
  • Keep the action in the verb! Makes the sentence stronger!
• Don't use the passive voice
  • Bio papers before 1950 used passive voice and were much harder to understand
  • In 1953 a classic paper was published that changed the writing style
  • Chris: "If you can add 'by zombies' at the end of your sentence, then you're using the passive voice."
    • "The object was moved (by zombies)." vs. "We moved the object."
  • The "Watson and Creek" paper
    • Google "rhetorical analysis of Watson and Creek paper"
    • The answer will be they changed writing by using active voice.
    • They used active voice to talk about their work and passive voice to talk about other's work
      • "We've shown X", "It was shown by Y that W is true"
  • "The oscilloscope displayed the voltage" vs "The voltage was displayed by the oscilloscope"
  • Not always possible to use active voice.
  • Try to keep it active voice, unless it gets boring. Must use passive voice too at certain points.
    • Q: Not clear how to alternate the voices?
• Correct word usage
  • Priciple vs. principal
  • Affect vs. effect
  • Alternative vs. alternate
  • Use a dictionary otherwise
• "Vigorous writing is concise"
  • The elements of style, by Strunk and White
  • Every word must be the right word, relate to its neighbor
• Write a story
• Use of abbreviations
  • i.e. -- that is (id est)
  • e.g. -- for example (exempli gratia)"
  • et al. -- and others (et alii)
• Spell checking
  • Will not always work "Roses smell sweet/sweat."

Illustration

Writing a research paper

• Hourglas diagram/model of Hill et al.
• Introduction is a trapezoid that gets narrower towards the end
  • It means you start out with a broad statement about your field
  • You want to bring the reader into your world; called the hook
  • the rest of the introduction is a focused discussion of the literature
  • the end talks about how your project will fill the gap in the literature
• Methods section
• Discussion section is a trapezoid that gets wider towards the end
  • begins with short summary of the results
  • rest is discussion of literature, except now you're saying "this is how my results address and extend our knowledge of what's already out there"
  • the end: why does this matter to the world
• Q: In Computer Science the organization differs
  • Intro presents problem, quick results and contributions and why it matters

Introduction

• A method to familiarize and orient
• Content depends on purpose of paper and the audience
• Written largely in the present tense
  • Could still have "X et al. showed that protein X is translocated to the nucleus"
• What's accepted in the field is in present
• Context: Orient your reader to the literature
• Focus: What questions are you addressing? What's your hypothesis.
• Justification: Show your work fits into and extends current work
• Common pitfalls:
  • Including unnecessary background, keep it focused (to what?)
  • Exaggerating or understating the importance of your work
  • Using lackluster openers (topic sentences must be strong) and weak follow-through in the body of your intro.
  • Including new results. (Q: In CS, contributions are stated in intro)
  • Improper tense (introduction is largely present tense)

Methods

• Describe the tools and processes
• Readers basically check your work in this section
• Pitfalls:
  • Providing too little or too much inforamtion
  • Reiterating published work rather than citing them
  • Writing strictly in chronological order
    • Most important first
    • Most fundamental first
  • Writing a protocol instead of a methods section
    • i.e. don't just write step by step what you did and nothing else
  • Putting too much detail in about stuff people in your field already know
    • e.g. like PK impersonation attacks?
• Usually written in past tense

Results

• Don't have a paper without results section
• Begin with an overview of the experimental design and stating the main findings to:
  • orient the reader
  • give the reader a sense of scope
• Report data in a logical order
• Results are primarily in the past tense

Discussion

This where you:

• Summarize findings
• Cite supporting literature
• Explain discrepancies between your work and prevent
  • Must emphasize what's new in your work
• You should point out shortcomings of your own work, not others, in this section.
• Define unsettled points.
• Discuss theoretical and practical implications of your work
• Why does this matter to the world.
• What did you expect to find and why?
• Don't be tempted by the squid technique: don't use ink just to obscure, make your words count
• What makes discussion tough?
  • Hard to define, requires perspective, knowledge, read the literature,
  • May have started before conclusions were formulated
  • May follow weak or inconclusive data
• Pitfalls:
  • Long wordy arguments that lack focus; lots of speculation
  • Failure to follow through with arguments from intro
  • Hedging excessively: wanna say something but you're not very sure of it
    • One hedging word per sentence like may/possibly/presumable/etc!
• Primarily written in present tense

Reference citations

The abstract and the title

Abstract

• "To abstract is to take away or remove the most important data from a set of records.", so it has to happen at the end.
• About 250 words
• Every abstract has a intro, methods, results, discussion and conclusion.
• Pitfalls:
  • All information in abstract needs to be in paper (no extra info)
  • Research paper abstracts are different than review papers
  • No abbreviations that are not described (unless DNA, RNA)
  • No scientific jargon? (Q: Whattt??)'
  • No references, no citations
  • Failure to state purpose

Titles

• Use minimal of technical terminology
• Use fewer than 80 characters
• Avoid active verbs, numerical values, abbreviations and punctuation
• No period at the end of the title
• Bad titles
  • Too wide
  • Too long
  • Ambiguity due to gerunds
  • Too general
  • Too specific
  • Too short
  • Incomprehensible

Always keep your reader in mind!