Tuesday, April 20, 2010

Research Update

Uncommon Ground has been hard at work this past month advancing our research on the relationship between grief and Facebook profiles, despite a recent lack of posts. After much deliberation, we have decided to narrow our research to focus specifically on analyzing the differences between language used in pre-death Facebook wall posts with that of post-death posts. Using pre-death posts as our control, we believe (and fervently hope) that the differences between pre- and post-death language use patterns which emerge from our analysis will ultimately shine light on motivations behind posting on deceased Facebook users’ walls. Here is where we currently stand in the research process:

THE DATA
We have collected data from the Facebook walls of twelve deceased Facebook users. Two sets of data exist for each profile: (1) all post-death wall posts, and (2) all pre-death wall posts within one year of the death. Items such as comments, notes, Facebook gifts, quizzes, status updates, and recent activity were deleted and not included in our data analysis as we are interested specifically in language use and not the use of other applications that the social network, Facebook, offers.

THE PREDICTIONS
In a rough sketch of our hypotheses, we defined several categories for comparative analysis, including, but not limited to:
  • complexity: we predict post-death wall posts will be more complex (i.e. higher word count, more six-letter words, longer sentences, more punctuation).
  • intimacy: we predict post-death wall posts will have greater intimacy.
  • positive & negative emotion: we predict post-death posts will use more positive emotion words.
  • verb tense: we predict more future tense usage in pre-death posts, and more past tense in post-death posts.
  • personal pronouns: we predict post-death posts to show a greater use of personal pronouns.
  • other…
  • religion: more religion-related language post-death
  • spacial terms: more spacial terms post-death
  • relative words: more relative words post-death

THE ANALYSES
We are currently in the process of analyzing our data using LIWC, WMATRIX and SPSS. Although we have begun to analyze our first profiles, we need to add results from more profiles to the mix before we can identify patterns.


Additionally, we are finessing the theoretical bases of our research and working on applying several theories (continuing bond, uses and gratifications, back stage/front stage emotion) to language analysis. We would highly appreciate any feedback/suggestions you may have on the theory-language connection or on our research in general.

Within the next week, Uncommon Ground will be working on analyzing all results to begin to identify patterns and form conclusions. Stay tuned for future posts on analyses and theories.

1 comment:

  1. There were some interesting topics brought up in the guest lecture from class today.
    Automated Sentiment Analysis:
    Understanding the tone of media coverage on a company or product. In our case it would be sentiment of Facebook users posting to deceased profiles.
    -using sentiment words
    -using sentiment relevant structures (e.g. negation, modals etc.)
    -using sentiment relevant feature (e.g. quotation marks for sarcasm, exclamation marks, etc.)

    All things that we are using LIWC to do in our data.

    ReplyDelete