[Sigcr-l] Questions for the year of cataloging research: #1 Does an LCSH subdivided heading have a meaning

[Simon Spero]

Here are some research questions that I think may be important

*Subject Headings (LCSH and others)*

Does a subdivided subject heading have a *meaning*? The results  in
Drabenstott (1998) appear to show that that roughly 50% of their chosen
subject headings could not be correctly paraphrased, even by technical
services librarians, let alone adult or children patrons.  The study has
methodological flaws, as discussed in the report itself.  The scoring
mechanism and sample selection may have introduced some bias factors.  The
paraphrase task may not have been an accurate measure of comprehension by
patrons, or composition by cataloguers.  Because pre-coordinated subject
heading strings are sequences of noun phrases, psycholingustic studies such
as Gleitman and Gleitman (1971) may be suggestive.
Unless speaker and hearer have the same understandings of the phrase
strings, then subdividing headings as currently practiced may be futile.

*Possible research:*

   1. Small rerun of Drabennstott (1998),  but stratify by education level
   per  Gleitman and Gleitman (1971) and use similar scoring approach.
   2. Use Drabenstott stimuli, with visual  presentatin, but with spoken
   responses.  Split modality allows the use of eye-tracking to reveal points
   of cognitive difficulty, or of unusual saccade patterns (e.g. the
   left-to-right phrases making up right-to-left  paraphrase may lead to a
   significantly higher level of microsaccades as the eye moves large distances
   in unusual directions.  See e.g. Reichle et. al (2004) for a good summary
   (It's in BBS, so it has the usual dozens of peer-responses and authors
   response).
   3. Set up online study using randomly selected headings stratified either
   by frequency of use of heading within corpus, frequency of use by
   circulation, or if available, frequency of use by global holdings.
   4. Select random sample of bibliographic records (SRS and stratified) and
   determine whether headings are subdivided so as to have their  formal
   interpretation appropriate to the work; to see  whether appropriate
   subdivisions could have been used by were not.
   5. Exploring of user understanding of headings occuring  in a task based
   context.  A forced choice experiment may also be informative.
   6. Other ideas?

*Potential Impact*

If results of Markey studied are validated for cataloguers and patrons, then
the entire use of subdivided headings is thrown in to question, and may not
be worth continuing for as long as subject headings are in string form.

If subdivided heading strings in selectively sampled records are
appropriately assigned, then the results for technical service subjects may
indicate distortion arising from the instrument or the paraphrase task.

If study users in a task based context show either correct or consistent
intepretation of subdivided heading strings, then the results for adults and
reference librarians may reflect issues in the original study methodology.

If eye-tracking shows increased cognitive burdens due to the structural
forms of the headings,  including unusual regression patterns, it may be
necessary to consider whether different presentation needs to be made of
such longer headings (including initially displaying headings in rolled-up
form, finding   more 4XXs for some subdivided headings, and also whether it
would be better to, e.g., move   $v of of the heading strings to reduce
length,etc .

Simon

Drabenstott, Karen Markey (1998).   Understanding subject headings in
library catalogs. Tech. rep. University of Michigan.
URL: http://deepblue.lib.umich.edu/handle/2027.42/57992
<http://deepblue.lib.umich.edu/handle/2027.42/57992%20>

Gleitman, Lila R and Henry Gleitman (1971). Phrase and paraphrase; some
innovative uses of language.
New York, Norton.

Reichle, E. D, K. Rayner, and A. Pollatsek (2004). “The EZ Reader model of
eye-movement control in reading: Comparisons to other models”. In:
Behavioral and Brain Sciences 26.04. Pp. 445–476.