Writing an Experiment Report

-taken from: http://www.cs.indiana.edu/~port/teach/541/expt.report.html-

L541, Introduction to Phonetics
R. Port, February 23, 2003

Writing an experiment report is a very stylized genre of writing. It should be as efficiently written as possible and should provide each item of information just where readers will expect it to be. Nothing redundant. The paper should contain 6 sections: a short Abstract, an Introduction, Methods, Results, Discussion and Conclusions. The Bibliography, Tables, Figures and typically come at the end (though the Tables and Figures can also be inserted within the text).

ABSTRACT:

Always include an abstract for ANY paper (in my opinion). Write it as one paragraph, about 200 words or less. This is a concise summary of the entire experiment. Include the rationale, method, results, and significance in highly abbreviated form (but using full sentences). Look at some published journal abstracts or Acoustical Society Meeting abstracts to get the idea of how to write one.

INTRODUCTION:

Give background rationale for the experiment. Why were you motivated to do this experiment? Or more accurately, given what you actually found, why MIGHT you have wanted to do this experiment? Your report, in other words, should not be biographical or historical. Experiments in phonetics are often exploratory; we often are not testing a specific hypothesis. And, in any case, readers really don't care what you THOUGHT you might find but did not. The paper should tell a story about what the data actually show. Summarize results of earlier research that are necessary to account for the conduct of the experiment (as relevant to the interesting aspects of your results) and help the reader to able to have some expectations about what the experiment will show (whether right or wrong).

Pay attention to your use of verb tenses in the text. It is easy (but confusing to readers) to slip back and forth between `The subject read the words at two speaking rates' and later `We measure the duration of the vowels by ...'. Avoid this.

The last paragraph before the Methods should sketch in readable style the basic logic of the experimental design to come. Eg, ``Thus, in the following experiment, Ss were asked to read words with both voicing values in position X at two speaking rates in order to see whether rate has an effect on the ratio of the vowel duration to consonant duration. We expected to find no difference in ratio due to speaking rate.'' (Or whatever.)

METHODS:

There are 3 logical parts to the methods: the Independent Variables (input variables, the things you controlled: place of artic, word length, speaking rate, etc or whatever), the Task performed by the subjects, and the Dependent Variables. You have a task (eg, reading these phrases in a some way) for which certain values of the IVs are set (eg, +/-voice, +/- front, final stops at two speaking rates, etc.). Then you measure the DVs (eg, vowel durations, F0 contours, whatever). Finally you look for ways in which the DVs are structured by (influenced by) the IVs. This section should be tight and business-like in style.

Of course, you need to tell your readers about at least the following:

• your word lists; number of repetitions; total number of tokens,
• characteristics of your subjects (sex, native language, age range, etc.; and the instructions you gave them),
• how the recordings were made, including major equipment items,
• analysis of the recordings; the measurement procedures; the software employed. Describe any special problems in measurement that arose and your solution to them,
• describe the statistics you performed, if any.

RESULTS:

Systematically go through the relevant aspects of your data. Present tables and graphs of the DVs. The organization of these paragraphs will probably be in terms of the IVs. For example, "The change in value of voicing affects both vowel duration and consonant closure duration," rather than "The vowel duration is affected by speaking rate, vowel identity and voicing".)

Describe the important features of the data in the text. Lead the reader through the tables and figures (eg, "As can be seen in the left side of Figure 1,..."). Devise data presentation techniques that make the real meaning of the data as clear as possible. Be sure to label graphs and figures clearly. Organization of paragraphs would normally be in terms of the independent variables.

DISCUSSION:

1) Organization of paragraphs (subsections) may often be in terms of independent variables (i.e. the factors varied in the experimental design).

2) Compare your results to previous similar experiments. To what extent do your results resemble those found before?

3) What new discoveries have you found? What is their significance to specific and more general issues in the area?

CONCLUSIONS:

Review and summarize very briefly what was done in this experiment and what you found in the results. State in general terms the most important discovery in the experiment. You may also want to suggest follow-up experiments.

Notes on Experimental Writing Style

1) Avoid first person pronouns no matter how awkward it may seem. "I asked the Ss to..." => "The Ss were asked to...".

2) Paper is cheap! It is easier to spread out your text and figures and use many separate pages.

3) Be as pithy as possible. Learn to use standard verbal formulas to abbreviate descriptions of methods, statistics, etc. Never be concerned that your paper is too short. It can only be too long. Say everything you have to say and no more. If you want to stretch out your paper, do it in the Discussion-- keep the Methods and Results tight, systematic and predictable.

4) Make a careful outline that you update as you write. It's generally easiest to write the Methods first, then Results (starting a new page), then sketch out the Discussion (starting a new page), and finally create an easily read and descriptive title. Then write the Introduction last.

5) When completed, the order of parts should be: Abstract, Text, References, Tables (on a separate pages unless very small), Figures (each on a separate page unless very small and labeled at the bottom). All tables and figures need a legend (on the page or listed on a separate page). Staple on diagonal in upper left hand corner. No fancy covers or plastic folders! Typed, of course, and EVERYTHING DOUBLE SPACED.

6) Use plenty of Figures! Be creative with figures to facilitate the reader's understanding.

But most of all HAVE FUN.