Making arguments

• Don't make evaluations; it's much better to describe a politically objectionable situation dispassionately and let your reader come to her/his own conclusions. However, feminist and queer traditions of activism in scholarship do allow for some leeway; you don't need to be disinterested, just careful and fair.
  Don't: "The ludicrously unrealistic representations of women in visual images in the media are absurdly obvious in language as well."
  Do: "Unrealistic representations of women in visual images in the media are obvious in language as well."
• However, you can be explicitly political in language, gender, and sexuality scholarship.
  Do: "Like other recent feminist scholarship on language, the present study supports the notion that gender is a performance achieved through discourse."
• Don't make grander claims than you can support empirically. Don't overgeneralize about "women" and/or "men" based on very small data sets. Say only what you can back up with data.
  Don't: "This study shows that women are more aggressive in conversation than men are."
  Do: "The women in this study were found to be more aggressive in conversation than the male participants were."
• Be cautious. Use hedges (perhaps, this suggests, women are said/claimed to... etc.) to make clear that you don't wholeheartedly endorse sweeping claims.
  Do: "Women are said to be more collaborative in interaction" (and cite some sources).
• Complicate the situation. Avoid oversimplifying complicated data. Instead, acknowledge the complexities, the limitations of your (or any) analysis, and go for the humbler goal of describing things tentatively, messily, and (more) accurately.
  Don't: "Our hypothesis was correct based on most of our data."
  Do: "Although our hypothesis was largely supported by the data, we also found some evidence to the contrary."
• Look for alternative interpretations. Your interpretation isn't the only possible one; acknowledge others and, if possible, show why yours is superior. But don't brush rival accounts under the rug.
• You can never prove anything through social science; you can only disprove things. Positive results can only support a hypothesis; they cannot prove it. In general, avoid the word prove.
  Don't: "Our findings prove that men talk more than women."
  Do: "Our findings support our hypothesis that in cross-gender interaction among young California adults, men produce more words and have more average floor time than women."

Using sources

• Make connections. Explicitly mention related studies, theories, etc. Whenever possible, include names of one or more relevant scholars. Every claim you make should ideally be documented or else demonstrated based on your own research. Social science often shows that what everyone thinks is obvious actually isn't, so don't assume that you can safely make assertions without backing them up.
  Do: "Gender-based teasing in childhood is a widespread phenomenon (Avery 1990, Barrett 1992, Jones 1995)."
• Don't make unwarranted attacks on previous scholars; they were doing the best they could. Be respectful and you'll be more likely to be listened to in the scholarly dialogue.
  Don't: "Smith (1972) naively assumes that women are generally introverted."
  Do: "Smith (1972) assumes that women are generally introverted, but this assumption is challenged by later research..."

Discussing data

• You're allowed to use the first person but don't get overly personal or autobiographical.
  Don't: "When I was a child, my brother used to tease me, and that's why I'm doing this research project."
  Do: "Gender-based teasing in childhood is a widespread phenomenon. This study considers the issue from a linguistic perspective..."
  OK: "Based on my own experience as well as previous research on the topic, I hypothesized that boys would be more likely to tease girls than boys."
• Don't hide. Acknowledge the role you played in the research situation; your presence certainly affected the results, so state that openly. If you were a participant in the data you analyze, it's best to refer to yourself in the first person.
• Make your speakers real people (even if they aren't). Don't just identify them by gender. Include details about other potentially relevant aspects of their identities: age, nationality, social class, race and/or ethnicity, religion, relationship to one another, etc. And try to think about how these other factors may be influencing their language use.
• Don't get into speakers' heads. We can't know people's psychological motivations or desires; we can only infer them from what they say and do and their effects on the interaction. Be especially wary about ascribing negative psychological states to speakers: low self-esteem, hostility, etc.
  Don't: "Susan wants to include her addressee."
  Do: "Susan's language use includes the addressee." (effect)
  Also OK: "Susan appears to want to include her addressee." (inference)
• Don't put interpretation into your transcript. This is where you present the facts, not your analysis.
  Don't: "Speaker (upset): I'm tired."
  Do: "Speaker (loudly): I'm tired."
• Integrate your data and analysis. The best papers don't present data in a list or appendix but present the data piece by piece and analyze each piece before moving on to the next. This makes it much easier to follow your argument. Don't make readers flip pages back and forth to follow your argument.
• Structure your paper as an argument, not a description. Don't present your data as a mere inventory of phenomena, but as evidence for a set of points you want to argue for. This often means moving data around, so that it's presented thematically rather than chronologically. Group data by what they illustrate, and then make clear what you're using them for. Don't' simply reiterate what transcript says in slightly different words; get beyond the words toyour offer an interpretation.

Format and style

• Make your analysis reader-friendly:
• Number each example and refer to numbers in your analysis.
• Likewise, number each line of transcripts of 4 or more lines and refer to the line numbers in your analysis.
• It also doesn't hurt to include precisely what text in each line you're focusing on:
  Don't: "Susan includes her addressee when she speaks."
  Do: "Susan includes her addressee in her statement through her pronoun choice in line 14: We can figure it out."
• Include transcription conventions in your paper (at the end or in a footnote), if you use any specialized symbols.
• When you discuss a word (phrase, etc.) as a word, italicize or underline it. (Quotation marks are okay but look a bit messy.)
• "This paper considers the gendered use of the discourse marker you know in conversation."
  Do: "The speaker's comment, I'm sort of tired (line 10), is hedged."
• Unless you have a good reason to do so, avoid the use of nonstandard spelling in transcripts. If phonetic detail is important, transcribe it using the phonetic alphabet or (in some cases) conventional colloquial spellings.
  Don't: "I wunna go cuz I'm sicka this."
  Do: "I want to go cause/because I'm sick of this."
• For quantitative analysis, include tables and charts in your writeup where each is relevant to the argument (as with transcripts), and refer to them explicitly. Explain to readers how to interpret the numbers in the tables.
• Illustrate your quantitative data by including examples of the categories you counted.
• Avoid empty words. Especially useless are interesting and intriguing. Instead, specify why it's interesting or intriguing.
• No future. By convention, academic writing doesn't use the future tense to describe its argument.
  Don't: "In this paper, I will examine..."
  Do: "In this paper, I examine...."
• Cite sources using the author-date system, which is conventional in the social sciences. Don't use footnotes; instead, provide a bibliography at the end of the paper, in which you list only those sources you cited in the paper.
• Make your paper look like a paper. This means double spacing (not one-and-a-half spacing!), providing a title, and numbering pages. It also means using a normal font size (12 point is standard) for text and data.
• Make your paper sound like a paper. This means including an introduction that sets the scene for your study, before you even mention your own research. Start with a broad question/issue/debate/problem, and make clear why it's important and interesting. Then discuss how your study addresses this issue. You should also have a conclusion stating the larger implications of your study for related research (and the larger world).