We need to get over the uncritical habit of regarding subject-heading strings--and the online public access catalog (OPAC) displays that bring them to our attention-as mere carryovers from the age of manual card catalogs. Instead, we need to consider them afresh in the light of their new and greatly increased power to aid researchers in the online age. By "strings" I am referring to the distinction between precoordinated, ordered phrases in Library of Congress Subject Headings (LCSH) versus postcoordinated Boolean combinations of individual terms, elements, or facets. For example:
Motion pictures for women-United States is a precoordinated string.
Motion pictures AND Women AND United States is a postcoordinated combination of individual terms in the form of a Boolean computer command. (Note that this combination also produces a meaning different from that of the precoordinated string.)
As a reference librarian who tries to pay some attention to cataloging literature, I've noticed a disturbing trend to regard providing web-type access to subject elements in the metadata fields of web records as more important than offering OPAC browse displays of LCSH in precoordinated strings. The problem is that search engines such as Google or Alltheweb or Teoma-or even OCLC's WorldCat-are incapable of providing such browse displays. The danger, then, is that a commendable desire to transform LCSH into a system more amenable for use in the web environment may simultaneously impair one of its major capabilities in the OPAC environment--one that is of critical importance to reference work.
Let me give one rather lengthy example: A reader came to me at the reference desk to ask for help in finding material on the history of Yugoslavia. He had searched our OPAC by doing a Boolean keyword combination of "Yugoslavia" and "history," and had been over-whelmed by the jumbled results, too many of which showed the right words indiscriminately in inappropriate contexts. To solve the problem, I showed him how to search using the OPAC's subject browse feature. By simply typing in the first word the researcher retrieved a list that included the following (as well as many more) precoordinated subject strings:
The reader was absolutely delighted with such an array of research options: It simply had not occurred to him that so many aspects of Yugoslavia's history could be so clearly distinguished or pursued in such detail. He did not, and could not, find this abundance of well-ordered material by keyword searching because it never occurred to him to use anything but "Yugoslavia" and "history" in a combination. Still, any or all of these displayed options (and dozens more not recorded here) could readily be of interest to someone researching the history of this country.
Even if this researcher had found the LCSH list-and he, like most others, did not know it exists-he still would not have found in the red books nearly the range of relevant research options spelled out for his perusal in the OPAC's browse display. The reason is that most of the strings displayed on the browse screens represent "free floater" subdivisions that are not identified in the LCSH roster itself. Moreover, the researcher also could not have found anything close to the same range of options by using tracings on the jumble of records that he did find via his initial Boolean combination, because they would not have alerted him to other facets such as "Bibliography," "Civilization," "Foreign economic relations," "Maps," "Social life and customs," and so on. (And in any event most readers don't know about the existence or function of tracings-the subject headings assigned to individual catalog records-any more than they know of the existence of the red books.)
The first crucial point is this: Even though all of these individual terms are indeed present in catalog records and findable through keyword combinations, most researchers will still never find them because it will simply not occur to them to enter such a variety of terms into Boolean combinations in the first place. …