17.442 gender-testing

From: Humanist Discussion Group (by way of Willard McCarty willard.mccarty@kcl.ac.uk)
Date: Wed Dec 10 2003 - 05:23:35 EST

  • Next message: Humanist Discussion Group (by way of Willard McCarty

                   Humanist Discussion Group, Vol. 17, No. 442.
           Centre for Computing in the Humanities, King's College London
                         Submit to: humanist@princeton.edu

       [1] From: lachance@origin.chass.utoronto.ca (Francois (98)
             Subject: From Linking to Gender Bending

       [2] From: "Prof. Shlomo Argamon" <argamon@iit.edu> (100)
             Subject: Info-age criticism and gender-testing

             Date: Wed, 10 Dec 2003 10:10:22 +0000
             From: lachance@origin.chass.utoronto.ca (Francois Lachance)
             Subject: From Linking to Gender Bending


    Elm has a wonderful feature... if I am too naysaying in a polemical reply
    or posting, I speed over the saving of the temporary file before the email
    is sent off thereby obliging me should I desire to reconsider a more
    constructive contribution to reconstruct the intended message. This is one
    such reconstruction. However it still seems a little tart in spots and I
    beg your indulgence and the forbearance of our readers especially in the
    passage below dealing with the rereporting of research results. [Note that
    my naysaying can of virago or of macho tone rarely in a hermaphroditic
    writing moment can I sustain the virulance of polemic. *wink*

    Dec. 1
    First there was Willard pointing to a (London) Times report on an article
    that appeared in Literary and Linguistic Computing. (The article residing
    behind a password protected site and I not being near a library with
    issues of the Journal at hand or license to access the online journal…)
    Only have Willard's excerpts to go on.

    Dec. 2
    Malcolm Hayward posts abstract from a paper he authored and the very
    significant observation that gender identification depends upon who is
    writing/speaking to whom about what (girls talking to girls about girls;
    boys talking to girls about girls, etc. etc.)

    Dec. 3
    Aimee Morrison raises, in passing, the issue of lay and expert audiences
    interpreting the results of gender-identification research.

    Stephen Clark raises further information about a particular case, although
    does not explicitly pose the question of collaboration: could two male
    authors collaborating together achieve a score for female authorship?

    Both postings call into question the sociological basis of gender
    ascriptions to the products of discursive behaviour. Is academic writing
    per se masculine? Is collaborative work per se feminine?

    Dec. 4
    Professor Argamon writes: "I am very interested in discussing this topic
    more - the links between lexicogrammar, gender, genre, and how text is
    perceived are very relevant, I believe, to developing an "information age
    criticism", bridging the gap
    between the "Two Cultures". "

    Dec. 5
    Francois Lachance reiterates Malcolm Hayward's call to be sensitive to
    context: "Who is a question that bears reflection in the context of where,
    when and
    with whom." As well, Lachance asks about ownership of the "Two Cultures"

    Dec. 7
    Patrick T Rourke points to C.P. Snow.
    Neven Jovanovic makes the plea for dialogue

    Dec. 7
    Lachance tries to be witty and ask a double question: who's or whose
    C.P.Snow? [The original posting that has thankful died in the ether had a
    long diatribe to the effect that Snow's formulation was itself routed in
    sexual politics and an ahistorical appropriation of it…. blah blah blah]
    Lachance does a search of the Humanist archive. Nets some very interesting
    exchanges on the culture bridging theme. Most interesting is this one…
    In that set of messages from Volume 3, Willard writes about the "eros of
    proof" and alludes to a passage in Homer's Odyssey. Methinks that the one
    who kicked off this thread tying us to the mast with a quasi-Siren* call
    should find the means and ways of retrieving the article in LLC and inform
    subscribers on if the article informs its readers on what grounds the
    ascriptions of gender were made. The tallying of the markers may be
    correct. Does the paper discuss the basis on which markers get to mark
    gender? It appears that neither the excerpts cited by Willard nor the
    summary that Willard gave to the Times piece give any indication that the
    interpretative basis was substantiated by verified scientific inquiry.
    Sampling. Sampling. Sampling.

    I am not asking for a regression to some putative foundation. I am asking
    for care in reporting. And even greater care in rereporting. Many of the
    messages on this thread, if accessed as single elements and not as a
    series, would lead a casual reader to the belief that the writers actually
    endorsed some of the under-specified premises about gender ascription
    (it's desirability and possibility).

    I find it very ironic that the thread on gender-testing was weaving its
    way into the Humanist archive in close temporal proximity of the linking
    anxiety thread where the distinction between argument and evidence was
    foregrounded. It strikes me that yes indeed "information age criticism"
    needs to bridge some gaps: between the question and that which provokes
    the question.

    Aside provoked by thinking of a culture of the question facing a culture
    of the assertion in a culture of the narrative: "What if the dimensions
    were not irremediably set in opposition? What if one considered sequence
    and figure to collaborate? One would face a machine. Every description
    as a state of being (configuration) possesses indexes translatable into
    questions for configuration's transformation (sequence). The nucleus of a
    narrative would be a description plus a question."

    * There are reading practices that let us entertain the Sirens and the
    Heroic Hero without identifying with either.

    O Helmsman! my helmsman our fearful trip continues

       -- Francois
    Lachance, Scholar-at-large http://www.chass.utoronto.ca/~lachance

    Wondering if...

              mnemonic is to analytic
              mimetic is to synthetic

             Date: Wed, 10 Dec 2003 10:14:47 +0000
             From: "Prof. Shlomo Argamon" <argamon@iit.edu>
             Subject: Info-age criticism and gender-testing

    From: lachance@origin.chass.utoronto.ca (Francois Lachance)
    > I wonder if Professor Argamon or others would care to elaborate on the
    > notion of an "information age criticism":
    > >
    > > Date: Thu, 04 Dec 2003 09:03:03 +0000
    > > From: "Prof. Shlomo Argamon" <argamon@iit.edu>
    > > [...]
    > >
    > > I am very interested in discussing this topic more - the links between
    > > lexicogrammar, gender, genre, and how text is perceived are very
    > relevant, I
    > > believe, to developing an "information age criticism",
    > bridging the gap
    > > between the "Two Cultures".
    > >
    > > -Shlomo-
    > Whose two cultures?

    As was already noted, this was a reference to C.P. Snow's notion of the Two
    Cultures of Science and Humanities. For an idea of what I meant, please see
    volume 2 (if I recall correctly) of Literary and Linguistic Computing from
    this past year (2003). When I have more time than I do now, I'll expand a
    bit more on my own vague ideas on how algorithmic text analysis might
    contribute to textual criticism.

    > Grammatical gender is marked differently in different
    > languages. And not all societies map social gender onto grammatical
    > gender.

    Quite right, which is why our work does not equate the two. I recommend (of
    course! :-) our papers on the topic, some of which are available at

    > Likewise perceptions and constructions of a science-art divide (I
    > think this is the intended reference of the two cultures) vary greatly
    > depending upon discursive contexts.

    True enough, but the distinction between "stripping away all meanings but
    the ones that everyone (rational) can agree on" and "generating meanings
    that connect disparate parts of existence for at least some people" is
    fairly strong, and makes the attempt to use algorithmics for deep humanities
    work rather difficult. (Even if my summary of the divide here is
    over-simplified itself.)

    > What does it say about the model at play when the researcher strips out
    > the count of pronouns and thereby ascribes the text to male-authorship?
    > What studies bolster the link between female-authorship and pronoun
    > peppering?

    This was not the issue here. Pronouns were ignored, and then all other
    features were re-evaluated, with the resultant new model evaluated again on
    the data, achieving only slightly less accuracy than the model with the

    > How extensive and valid is the empirical data that informs the
    > assertions made by the model?

    Our study covered published work in modern British English taken from the
    BNC. The corpus was, of necessity, not as large as I would have liked, but
    it was large enough to evidently glean some reasonable (and likely robust)

    > The sampling on gender-based linguistic
    > analysis needs to account for such factors as socio-economic status, age,
    > first-language, inheritance of language patterns from gender of the
    > primary language teacher, production context of the utterances (e.g.
    > gender of the interlocutors). If the model can begin to account for the
    > sociolects and the ideolects of speakers and writers, the number crunching
    > analyses might be put to good use in terms of gender-testing not only the
    > sociosexualization of the writers but also that of the intended readers:
    > M to M
    > M to F
    > F to M
    > F to F

    One of the points of our study was to show that even with a very amorphous
    "intended reader", sex-linked language effects show up. I would *love* to
    have a large corpus that was stratified in all the ways that you suggest.
    Heck, give me a million dollars, and I'll build one myself! (Corpus
    construction is, of course, a major bottleneck here.)

    > And that whole dynamic shifts when moving from a one-to-one to a
    > one-to-many:
    > M to M&M
    > M to M&F
    > When the phenomenon of split addresse is observed and taken into account
    > the unicity, especially the gendered uniformity of the author may no
    > longer cohere.

    If I understand your point here correctly, I must disagree, since we showed
    that the gendered identity of the author does in fact inhere in texts
    published for a wide audience, in a variety of genders. Please read our
    papers; I would be most interested in your comments thereon.

    > Who is a question that bears reflection in the context of where, when and
    > with whom.

    This is nearly tautological these days, but the empirical evidence does not
    quite fit. Perhaps authors of different genders have different (composite,
    vague, amorphous) imagined readers in mind, even when writing for a broad
    audience? I don't know, and it would not be that easy to really find out, I


    Shlomo Argamon, Associate Professor
    Department of Computer Science, Illinois Institute of Technology
    Chicago, IL 60616
    Phone: 312-567-5289 Fax: 312-567-5067

    This archive was generated by hypermail 2b30 : Wed Dec 10 2003 - 05:28:04 EST