Chapter 2
Subject Headings and the Library Catalog
Why should you use a library catalog in addition to the Internet? There are two major reasons:
WHAT ONLINE PUBLIC ACCESS CATALOGS (OPACS) CONTAIN
A library’s online catalog primarily lists the book holdings of the institution; in current terminology it is often referred to as an online public access catalog (OPAC). Your local OPAC may also list nonbook formats such as manuscripts, sound recordings, videos, maps, photographs, and so on; it may also provide links to e-books that your library offers. The titles of journals and magazines (e.g., Harvard Business Review, Sports Illustrated) held in physical formats (paper or microfilm) by the library will also be recorded, but, as a general rule, the catalog will not enable you to search for the individual articles that appear within the journals. (For those you will need separate databases and indexes, discussed in Chapters 4 and 5. For determining the library’s holdings of journals in electronic formats, see Chapter 4.) Each entry in the catalog will provide you with a call number enabling you to locate the desired volume on the shelves.
PROBLEMS IN DETERMINING THE RIGHT SUBJECT HEADINGS
The most frequent, and most serious, problem that people have in looking for books in the catalog is the determination of the right subject headings to use. For example, if a reader wants books on morality, should she look under “Morality” or under “Ethics”? Or must she try both? Or perhaps, with a more focused topic in mind, she should look directly under “Ethical relativism”? Similarly, if another reader wants information on sentencing criminals to death, should he look under “Death penalty” or under “Capital Punishment”? And how does he know that he’s thought of all of the right terms? Perhaps he should look under “Execution” or “Lethal injection” or some other terms as well.
Reference librarians routinely deal with people having problems in this regard. One student, for example, became frustrated in looking for material under “Moonshining” because it is not entered under that heading. In a standard library catalog, works on this subject are recorded under Distilling, illicit. The student who was researching the Huron Indians found, upon asking for help, that the preferred subject term is actually Wyandot Indians. (Similarly, the proper OPAC heading for Chippewa Indians is Ojibwa Indians.) Another researcher wanted books on “Corporate philanthropy”; before talking to a librarian she hadn’t found anything on target because she was looking under “philanthropy” rather than under the proper heading Corporations—Charitable contributions. Researchers who want “Multinational corporations” often make the mistake of searching under that term when the proper heading is actually International business enterprises; those searching for “Test tube babies” usually fail to search under the proper heading, Fertilization in vitro, Human.
Not only the choice of words but also their order may be confusing—for example, should one look under “Surgical diagnosis” or “Diagnosis, surgical”? under “Heavy minerals” or “Minerals, heavy”? under “Fraudulent advertising” or “Advertising, fraudulent”? Inverted forms are not used consistently, so there is much room for error unless you catch on to the easy systems for getting to the right choice. (Some OPACs solve these problems by searching for terms in any order; some do not.)
LIBRARY OF CONGRESS SUBJECT HEADINGS
You do not have to just guess—there are indeed systematic steps to take that will solve most of these problems. Specifically, there are five ways to find the right subject headings for your topic—to get from the terms you think of to the often different terms used for categorization purposes by the library catalog. Two of these ways are available through an annually revised online list called Library of Congress Subject Headings (LCSH) published by the Library of Congress; as of July 2013, it is available as a free PDF online: www.loc.gov/aba/publications/FreeLCSH/freelcsh.html. This is the standardized list of terms and phrases used by libraries throughout the English-speaking world for categorizing the books in their collections. All college, university, and research libraries use this list; so do most large public libraries. (Small neighborhood libraries may use a different one called the Sears List of Subject Headings; this is a much shorter one-volume roster that does not make all of the fine distinctions that are needed for categorizing really large collections.)
There is also a printed version of the LCSH in a 6-volume 35th edition (2013) set; it is often referred to simply as “the red books” because bright red was the color of its binding. Since 2013, however, only the online PDF version has been available, but the printed edition will still be useful for several more years, and its format makes it easier to flip through.
Apart from consulting the LCSH list directly (whether PDF or red books), there are three other ways to find the best subject terms for your topic just by using mechanisms within the library’s online catalog itself. It must be noted, however, that many libraries have recently opted for Internet-type single search boxes that do not provide the mechanisms of access described below, a point to which I’ll return. In such situations you can still use the LCSH system via the online catalog of the Library of Congress at catalog.loc.gov for subject access, and then check the titles discovered there to see if your local library owns them.
Before looking at all of the five ways to get from your keywords to the right LCSH category terms, however, it is important to consider the principles governing the compilation and use of the Library of Congress Subject Headings system, because these principles put catalogs into a different universe of search capabilities—more predictable and more systematic—than the realm of the open Internet. They are:
These may sound like deadly dry irrelevancies to anyone who wants to get in and get out quickly in doing research, but an understanding of how these principles structure the catalog will actually give you the best means of doing just that—of using the OPAC quickly, with maximum efficiency, and without any wasted effort in having to sort through mounds of irrelevant or tangential hits (produced by keyword searches) that don’t really provide exactly what you want. It is the combination of these three principles that enables the library OPAC to show you “the shape of the elephant” of the book literature on your topic—i.e., enabling you to see all of the relevant titles in your library (not just those whose keywords you can guess) and their relationships to each other, in ways that weed out hundreds of irrelevant works having the right keywords in the wrong contexts and that give you reasonable confidence you haven’t missed something important.
UNIFORM HEADING: STANDARDIZATION OF SEARCH TERMS
“Uniform heading” is the principle that addresses the problem of synonyms, variant phrases, and different language terms being used to express the same concept (e.g., “Death penalty,” “Capital punishment,” or “Todesstrafe”; “Flying saucers,” “Unidentified flying objects,” or “UFOs”). Authors who write about any particular subject simply do not use identical terms to refer to it. Librarians who create systematic catalogs, however, solve the problem by choosing one of the many possible terms, in such cases, and entering all relevant records under that single category term, rather than repeating the same list of works under each of several terms. Since the full list of relevant books (or other records) appears under only one of the terms, the catalogers will insert cross-references from several of the other possible terms to steer readers to the one main grouping.
For example, books on the subject of the Cockney dialect have titles such as the following:
No one researching this topic would be able think up all of these unusual keywords if confronted only by a blank search box, as in Web search engines. Library OPACs, however, round up all such books because the library’s catalogers have artificially attached to each record the same (uniform) heading, English language—Dialects—England—London. When this cataloger-created heading is affixed to each different title record, the result is that all of those records then have a retrievable point of commonality that would not otherwise be present. The researcher who finds just this one common subject heading, then, can thereby round up all of the disparate title records to which it is attached, without having to guess all of their variant title (or other) keywords.
Similar scattering shows up in just about any other subject area. Here, for example, is only a brief sampling of the many variant book titles that are cataloged under the heading Capital punishment:
Hundreds of other keyword-variant titles could be listed—and in scores of languages other than English. Without the creation and assignment of the artificial point of commonality (the LCSH term Capital punishment) to each record, a researcher looking for what the library has to offer on this topic would miss most of these works. Their own keyword terms are simply too diverse to be rounded up systematically by computer algorithms (which usually rank the displays of only the exact words that have been typed in, without noticing synonyms or variant phrases for the same concept); nor could the full variety of the books’ own keywords be anticipated even by researchers having some prior expertise in the subject area.
This is an important point: relevance ranking of keywords by Internet search mechanisms is very different from conceptual categorization brought about in OPACs. The former may massage the display order of records containing the words you’ve typed in, but it won’t bring up records having any of the other ways of articulating the same idea. (There are minor exceptions here, in that some search engines do automatic word stemming—which means that, for example, entering “homeowner” will also retrieve variant endings of the same word stem such as “homeowners” or “homeowner’s.”) Projected large-scale semantic Web solutions, in which computer algorithms attempt to aggregate synonyms or variant phrasings, don’t work now even within English, let alone across multiple languages simultaneously. It is therefore prudent to accept only with a grain of salt claims for their future success. Moreover, the synonym linkages in semantic webs are such that you can never tell which links have not been made—look again at the unpredictable variety of title keywords in the examples above. Subject headings, in contrast to keywords, are conceptual category terms—if you find the right LCSH heading, you’ve found essentially all of the books in that category, no matter what unpredictable terms the books may use themselves. (I’ll have to qualify that word “all” below.)
DISPLAYING UNANTICIPATED ASPECTS AND RELATIONSHIPS: BROWSE MENUS
A uniform heading also serves to round up the different aspects of a subject through the use of subdivisions of the lead term in the verbal string. (This is an area in which LCSH differs in an important way from other thesauri, or lists of controlled indexing terms that are used in other databases.) In the “Cockney” case, the proper heading not only indicates the subject of the books to which it is assigned, it also appears in the library catalog’s browse display in a way that shows how it is related to other aspects of the topic. The following is only a brief sample of the hundreds of subdivisions that show up under English language—Dialects:
Sometimes people wonder why a complicated heading like English language—Dialects—England—London is used rather that more obvious and direct terms such as “Cockney” or “Cockney dialect.” The latter forms are certainly preferable by a simple criterion of popular usage, but there’s more involved in the cataloging system than just that one consideration. The reason for using the string is that, as in this example, it is crafted to draw attention to a larger context of relationships to other relevant subject headings. In this case, a researcher interested in the Cockney dialect may well discover, through such a browse display, several other aspects of English dialectology that might be of interest.
Browse displays of the many aspects of a topic enable researchers to recognize what they cannot or did not specify in advance; the larger a library’s collection, the more readers need such menus to serve as road maps into the range of available, but unexpectedly relevant, resources. (Note further that semantic webs may succeed in rounding up some synonyms and variant phrasings for an idea—in one language—but they cannot bring to your attention related ideas that are “off to the side” and that alert you to the larger surrounding contexts of a subject. This is not to say that semantic webs are useless—far from it; it is simply to point out that there are real trade-offs involved, and that they are no substitutes for the LCSH system.)
The creation of browse displays of headings with subdivisions (as above) is one of the main differences between a library catalog and a mere inventory of holdings; the latter will record individual items but will not display subject categorizations of the items or show relational linkages among the categories themselves. Inventories give you lists of the items within a collection; library catalogs provide not just lists of the items themselves but higher-level subject linkages: the subject headings themselves round up keyword-variant texts on the same subject so that the variants can be seen in relationship to each other; and both cross-references and browse menus systematically map out the further, “outside” relationships of the subject categories themselves.
Library catalogers who create headings have to be mindful not just of the appropriateness of any individual term (that “artificial point of commonality”) that they assign to an individual catalog record (for a book) to designate its conceptual grouping, but also of the “browse” position in which the chosen form of heading will appear in relation to other established headings.
Differences from Tags
This concern also points up a major difference between formal LCSH headings and keyword “tags” added to records by amateur enthusiasts, as in LibraryThing.com—the latter index terms are words attached to individual book records without any regard for the uniform application of the same term(s) to all of the other relevant books on the same subject. In other words, the consistency of tags from one record to the next is itself as unpredictable as the range of keywords in the books’ titles. Tags are assigned by readers’ whims, without regard to any cataloging principle of uniform heading. Moreover, tags are applied without any intention of placing them in a formal network of relationships that will display other concepts that may also be of interest, but which are a bit “off to the side” of the one particular book in hand—i.e., the one being tagged. Searching on the tag “Cockney,” for example, will not bring to your attention other works on different dialects, such as English language—Dialects—England—Dorset.
LINKAGES TO CLASSIFICATION NUMBERS
The creation of uniform terms with standardized subdivisions serves yet another purpose in relating a library’s books to each other. The many possible subdivisions of a single term (e.g., Bibliography, History, Law and legislation, Study and teaching) may serve to direct researchers to entirely different call number areas in the bookstacks, scattered throughout the classification scheme.
For example, works under the LCSH term Small business usually get classified and shelved in HD (Economics) or KF (U.S. law) areas, but if the books being cataloged are assigned the Accounting subdivision, they switch to HF (Commerce) classes instead. If they receive the Finance subdivision, they are usually classed in HG (Finance); if they are on the bibliographical aspect of the topic (Small business—Bibliography), they are classed in Z7164.C81 (Business bibliography). All of these different class designations, while scattered in the bookstacks, nevertheless appear together in the catalog under the one heading Small business (with its various subdivisions). Thus, catalogers who choose one heading form over another have to be thinking not just of the conceptual relevance of the verbal string to the book in hand, and not just of its contextual display in a browse list, but also of which classification number may be tied to the string that is created. (Further, they have to keep in mind the heading’s position in a network of broader, related, and narrower subject relationships defined by cross-references.)
SUBJECT HEADINGS IN THE OPAC AS INDEX TO THE CLASSIFICATION SCHEME
There is not a one-to-one relationship between all LCSH headings and specific classification numbers, but there are so many tens of thousands of such formal linkages that the subject headings in the catalog effectively function as the index to the classification scheme in the bookstacks (see Chapter 3). Finding even slightly different subject headings in the OPAC may bring up catalog records whose call numbers point you to entirely different areas of the bookstacks if you want to browse the shelves. Good catalogers are fully aware of this linkage—which provides yet another important distinction between a library catalog and a simple inventory list. The same consideration of linkages to classification numbers also points up another substantive difference between Library of Congress Subject Headings and “descriptors” in conventional subject thesauri—the latter terms do not need to be linked to class numbers in a library shelving scheme. This point also applies to the application of keyword tags, by amateur contributors, to book records in LibraryThing (and elsewhere)—those tags are assigned without any regard to, or awareness of, appropriate or consistent call number linkages defining where the books will be physically shelved in relation to other books. For example, a large library cannot sanely assign all of the books on the various aspects of the subject Afghanistan to a single classification number; there are too many distinctions that have to be made within that large topic, without which researchers would be overwhelmed by too many irrelevancies. Distinctions must be made both within the LCSH headings themselves (via subdivisions) and within the class numbers that determine where the books will be shelved. Thus, specifically distinguished aspects of the subject heading in the catalog are also frequently tied to specifically distinguished classification numbers:
Without these distinctions in the OPAC subject headings, the 10,000+ books on Afghanistan at the Library of Congress would all be jumbled together incomprehensibly in the bookstacks in a single undifferentiated mass. (A similar, although lesser, problem would exist in all other research libraries. The importance of shelf-browsing capabilities within different aspects of a large subject will be discussed in Chapter 3.) Note that these class distinctions could not be designated in the first place without the corresponding LCSH headings being created as multiword strings—a feature of LCSH headings that make them unlike the single-term descriptors used in conventional thesauri. Such strings are entirely eliminated in the faceted OPACs that are now, unfortunately, used by many university libraries. (A faceted catalog breaks up subject strings into their individual component words. See the discussion of precoordination and postcoordination, below.)
Uniform headings in OPACs thus round up in one place both variant titles for the same subject that may be alphabetically scattered among keywords from A to Z and variant classification numbers for aspects of the same subject that are scattered throughout the bookstacks.
RECOGNITION ACCESS (PROVIDED BY CONCEPTUAL CATEGORIZATION RATHER THAN RELEVANCE RANKING)
The emphasis in library catalogs on collocation via uniform subject headings points up a major problem with Internet search engines, in contrast: no matter how sophisticated their relevance ranking algorithms may be, they are still ranking only the keywords that you’ve typed in to begin with. (Even if, in some cases, there are under-the-hood links to other terms, you still cannot tell which terms are linked and which are overlooked.) If those terms are not the best ones to use in the first place, manipulations of their rank-ordering will do nothing to bring about the retrieval of entirely different words for the same subject—i.e., typing in “Cockney” will not retrieve “Fraffly” or “Muvver” or “Vulgärsprache,” as in the above example. (Nor will it bring to your attention the many other geographical options arrayed under English language—Dialects.) Typing in “death penalty” will not retrieve “ultimate coercive sanction” or “murtherers” or “unforgiven”—nor will any semantic web make such connections automatically. Nor will user-assigned tags. Relevance ranking, again, is not the same as conceptual categorization. The latter function is absent in Web searches. Internet or Amazon-type searches may provide some linkages of a very different—and often very useful—kind (e.g., “Customers who bought this book also bought these”), but that sort of linkage cannot find conceptually similar works, with variant keywords, in a systematic manner; and it usually does not notice relevant foreign-language or out-of-print works at all. Library of Congress Subject Headings, in contrast, round up all (or at least most) of the books in a library’s collection on a given topic, no matter what keywords their authors used, in either English or foreign languages, and both in print and out of print—they all show up under the one uniform heading chosen by the catalogers. Further, all of the library’s books—which have been professionally selected to begin with on the basis of quality—will appear there, even if no customer/user has yet read or recommended them.
There is thus a huge difference between online library catalogs and Internet search engines or Amazon displays. The former, constructed on the principle of uniform heading, enable you to recognize, within a retrieved subject set, a whole host of relevant titles whose variant phrasings—such as those above—you could never have specified in advance. The crucial element of serendipity or recognition at the retrieval end of library searches is a direct function of keyword-transcendent categorizations having been created by librarians at the input (cataloging) end of the operation.
Cataloging is thus not at all the same as merely transcribing existing data from title pages or tables of contents (as in an inventory), nor is it simply a matter of adding keyword tags that are not uniformly assigned to all of the conceptually relevant records (as in LibraryThing). It is a process of adding terms that are standardized “on top of,” or in addition to, the words provided by the book itself, or by tags contributed unsystematically by amateurs. These standardized elements serve to collocate under one subject heading the widely varying expressions used by many different authors, worldwide, in talking about the same subject; and at the same time they function to bring to your attention a whole network of cross-references and menus of related aspects of your subject, off to the side, that you probably wouldn’t think of on your own until you saw them listed for your inspection. Moreover, these category terms themselves, unlike title (or other) keywords or tags, can be identified without guesswork through predictable and systematic means (the five methods discussed below).
SCOPE-MATCH Specificity and Its Modifications
“Scope-match specificity” is the second principle governing the operation of Library of Congress Subject Headings. Its meaning requires some understanding of library history. Prior to the advent of computer catalogs, books as a general rule were seldom assigned more than two or three standardized subject headings. This limitation was an important consideration in the era of card catalogs, because filing individual cards for each book under a half dozen or more terms at different places in the alphabet would result in a catalog that was physically very bulky. The fewer categories to which any book was assigned, in other words, the fewer cards had to be filed, and the more manageable was the size of the overall physical file—a matter of considerable importance in research libraries holding millions of books.
Another important consideration—still relevant even in the age of computer catalogs—is the volume of work that catalogers have to do. Although the number of books published every year is huge, almost all libraries are chronically underfunded, and so catalogers who have to create records for dozens of books each day don’t have the time to figure out 10 or 12 headings for each one. Again, cataloging is not simply a matter of transcribing words from titles or tables of contents—nor is it simply a matter of employing existing headings from the LCSH list. It is also a matter of extending the list, creating new headings, and—the difficult part—integrating all of the new terms into an intricate webs of cross-references, which require the specification of hierarchical relationships (broader, related, narrower) to many other headings. Considerations of the best form of heading to create—and how that form will show up in relation to other terms in browse displays—are also important, as are considerations of how new headings may be linked to new classification numbers, which may also have to be created at the same time. The intellectual work involved, when done well, is quite intricate and challenging. It’s sort of like doing multidimensional crossword puzzles—you have to get words that are not just appropriate to a particular book but that also fit into several large patterns of linkages to other terms that already exist in the same system.
Considerations such as these affect scope-match specificity. In the card catalog era, the principle meant that catalogers would usually assign the minimum number of headings that indicated the subject content of a book as a whole—that is, catalogers would not assign a standardized term for each individual chapter or section of the book. Thus a book about Oranges and Grapefruit would have both terms assigned to its catalog record, because in combination they covered the whole scope of the work. If another book dealt with these two fruits plus Limes, a third subject heading would be assigned. However, if a book dealt with Oranges, Grapefruit, Limes, and Lemons, in traditional cataloging it would not receive all four headings; rather, a single generic heading representing all of the subtopics comprehensively would be assigned—in this case, Citrus fruits. While this is indeed a generic term, it is nevertheless the most specific one that covers the book as a whole. Catalogers traditionally aimed for this level of coverage in the card catalog era. If there was not a single term that expressed the subject of the book as a whole, the goal was to sum up the book in as few headings as possible—usually about three.
The advent of computer catalogs more than a generation ago eliminated the need to worry about physically bulky card catalogs bursting with too many cards in too little space to contain them, so computerized catalog records often receive more than three subject headings. Records created in earlier decades, however—especially prior to the 1980s—still reside in online catalogs with what today would be considered nearly minimal subject headings on them. So, if you need to search for early books, remember that you do not have as much leeway in your choice of headings as you may have in searching for more recent books.
The welcome development of computer catalogs—and especially of full texts of books appearing online—has led some commentators to assert that subject headings are no longer necessary at all; they say, in essence, “Why is there a need for adding any terms to catalog records when so many more searchable words have now become available through full-text search capabilities?” One major answer to this question lies in the fact that scope-match specificity in subject headings solves one of the biggest problems that researchers have in trying to find books relevant to their interests. (The other answers entail an understanding of the principles of uniform heading and specific entry; but the people who ask the question to begin with usually don’t want to hear the replies.) As anyone who has ever used Web search mechanisms discovers immediately, typing keywords into a blank search box will overwhelm you with thousands, if not hundreds of thousands, of websites or e-books having those words anywhere in their texts.
Solving the Problem of Excessive Granularity
What researchers really want, especially when they are at the initial stages of an inquiry, is not a roster of every book having the right words appearing in any context, no matter how tangentially. What they want to start with is a much smaller list of whole books on the topic—or at least only those books having substantial sections on the desired subject. They want first to find the substantive treatments of their topic, not just tangential references to it. This is exactly what scope-match subject cataloging provides immediately and that keyword searching of full texts misses. Full-text keyword searching usually provides results that are much too granular—that is, it buries the substantive works within thousands of essentially irrelevant retrievals. (There are of course other times when very granular retrieval is highly desirable; see Chapter 5.) You cannot see “the shape of the elephant” of the book literature on a topic by doing full-text keyword searches; you will get way too many hits that have the right words in the wrong contexts, or that have only tiny sections pertinent to what you want. And the presence of so many thousands of irrelevancies serves to hide from view many of the best sources. Scope-match searching (via LCSH headings), in contrast, retrieves whole books on the desired subject, segregated from thousands of irrelevant retrievals.
It is perhaps ironic, but it is nonetheless true: a cataloging rule that was created to solve a different problem in the card catalog era (that of preventing catalogs from becoming too physically bulky) now serves to directly solve one of the most serious problems of the online age: that of excessively granular retrievals—i.e., way too many irrelevant works being brought up by full-text search capabilities. Researchers who understand the advantage of scope-match subject cataloging in OPACs thus have a major advantage over those who cannot see retrieval possibilities beyond keyword searching of full texts. Cataloging retrieval will bring up not just conceptually related works (no matter what keywords or languages their authors have used), it will also bring up only the most substantive relevant works, freed from burials within thousands or millions of irrelevancies.
The current Subject Cataloging Manual of the Library of Congress, which is more or less the standard for catalogers in all large libraries in the English-speaking world, allows for a “20 percent rule”—that is, a heading may be assigned for any topic that takes up 20 percent of a book’s content. The manual further notes that “generally a maximum of six [headings] is appropriate. In special situations more headings may be required.”
Problems with Copy Cataloging
I say the Manual is “more or less” the standard because many libraries today have chosen to produce records at a less rigorous level, called “core cataloging” or “cataloging on receipt” (COR). At this “core” level, two subject headings per book are usually deemed appropriate.1 Other books may receive “minimal-level cataloging,” which means that their titles are recorded but they are assigned no standardized subject headings at all. These lower levels of work are turned out by some libraries because of the huge volume of book publication coupled with the decreasing number of trained catalogers. Indeed, traditional cataloging is a subject no longer required in some library schools. Additionally, most libraries do “copy cataloging” as much as possible, which means that they strive to find within the OCLC system catalog records already created by some other library—any other library—and simply import them into their own catalogs, frequently with minimal review or quality control of the copied subject terms. (OCLC used to stand for “Online Computer Library Center,” a network incorporating the catalogs of more than 70,000 libraries worldwide; now it’s just plain OCLC.) Obviously the quality of cataloging contributed from so many sources varies a great deal—which causes the principle of uniform heading to be considerably undermined in all too many instances. This is a reality that researchers have to take into account: You do want to start by looking for what are presumably uniform headings, but you also have to know how to navigate through the unavoidable inconsistencies that are unfortunately becoming more common.
Here’s an example of the problems that the “core” standard can create. A book published 20 years ago, in 1984, with the title Censorship and Interpretation: The Conditions of Writing and Reading in Early Modern England, received a generous set of headings:
A similar book published in 2001, however, Press Censorship in Jacobean England, received only two headings:
There are a couple of problems here. The first is that the person who cataloged the second work lacked sufficient subject expertise and mistakenly assigned “16th century” rather than “17th” for the Jacobean period. And here it must be admitted that library work is as much an art as a science; despite its many rules designed to bring about uniformity and standardization, they are only as good as the person using them—and good subject cataloging requires some expertise in the subject area as well as a knowledge of the cataloging rules.
The more serious problem is that the first book, Censorship and Interpretation, is widely considered the standard work in its field—but a researcher who looked in a library catalog under any of the headings used for that book would not have found the later 2001 work, which covers much of the same ground from a different perspective. The minimal two headings given to the later volume were simply not adequate to place it into any of the same categories in which the first book appears—and it should show up in some of those same groupings. The problem in this case was solved by a “guerrilla cataloger” who, once the difficulty was pointed out, simply ignored the core cataloging standard and added to the 2001 record several more of the LCSH terms that now enable researchers to find the two books under some of the same headings.
Not all such problems, however, can be solved by guerrilla catalogers who will ignore the rules they are told to work with and do better work than the core standard requires. This means, again, that the “uniform” part of the principle of uniform heading has been considerably diluted in the library profession in recent decades, and so researchers need to be aware of the ways to compensate for decreasing standardization.
Specific Entry
This is by far the most important principle that researchers need to be aware of. It means that, given a choice between using specific or general headings for a book, catalogers will predictably choose the most specific possible heading(s) for the book as a whole, rather than the more general headings that are also available in the LCSH list.
For example, if you are looking for material on nightmares, you should not look first under Dreams or Sleep but under Nightmares specifically. Similarly, if you want books on Siamese cats, you should look under the specific heading Siamese cat and not under the general heading Cats. One researcher looking for material on Jewish children mistakenly assumed the proper heading would be Jews. It isn’t. It’s Jewish children. Works under this more specific heading are not also listed under the more general heading Jews.
Another reader looking for information on recreation rooms searched under Homes. The right heading is Recreation rooms.
A reporter looking for material on game shows looked under Television. The right heading is Game shows.
A reader looking for information on archons (chief magistrates) in ancient Greece searched under both the keyword phrase “ancient civilizations” and the heading Civilization, ancient. The right heading is Archons.
A reader looking for information on emblem books searched under Iconography. The right heading is Emblem books.
A reader looking for information on the problems that alcoholic parents create for their children wasted a lot of time after being inundated with records under Alcoholics. The right heading is Children of alcoholics, and there is also an even more specific term, Adult children of alcoholics. Again, books listed under the narrower terms are not also listed under the general heading—you have to find the most specific headings rather than (not in addition to) the general terms. Still another reader interested in the effects of divorce on children made the usual mistake of looking under the general heading Divorce rather than under the specific heading Children of divorced parents. (And there is also a heading for Adult children of divorced parents.) The materials listed under narrower terms do not also appear under the broader term or terms; the choice of which term(s) to use, when several levels of generality are possible, is predictably made on the basis of the principle of specific entry.
This is not to say that the general headings aren’t used—of course they are. If a book is about Divorce in general, then it will receive that heading, because in that case the general heading is indeed the tightest fit for the scope of that book.
The problem is that most people search under general headings when they really have something more specific in mind; they simply assume mistakenly that the general categories “include” the more specific topics. The rule for researchers, then, is to start with the most specific terms from the LCSH list and then “go general” only if you cannot find narrow terms that match your specific topic exactly.
Note that this rule runs counterintuitively against the grain of what students are usually told: “Start with a broad idea and then try to narrow it down as you go.” While this may be very good advice in other situations, you should do the exact opposite when you are using a library catalog. (I dearly wish someone had pointed this out to me when I was still a student myself.)
The problem for researchers who start with general terms is that in doing so they usually do find a few sources that appear to be in the ballpark—but they simultaneously miss most of the best material without knowing it, and they usually stop with their initial pool of general sources. The researcher who looks under Divorce, for instance, may indeed find a few books that have sections discussing the effects of divorce on children—but if she stops there (and most people, like the Six Blind Men, do stop at the first level that seems at all relevant), she will miss all of the works under Children of divorced parents, which are whole books (rather than just chapters or section) on that topic. Progressively refining the wrong initial search set by adding extra keywords will not magically transform it into the right initial search set.
Preventing Overload in Coverage
There are two good reasons for the specific entry rule. First, the segregation of different levels of specificity prevents any of them, broad or narrow, from becoming overloaded with irrelevant clutter. For example, the heading Decoration and ornament is quite general, but linked to it (via cross-references) are 125 narrower terms such as:
If all of these narrower terms were included within the scope of the general heading, then researchers who wanted only the general-level books on Decoration and ornament would not be able to find them without having to wade through, at the same time, all of the hundreds of cluttering records for the much narrower topics. Specific entry prevents general terms from being overloaded by the inclusion of too many related but narrower topics, and sometimes researchers do want to start out with very general “overview” books.
Predictability in Selection of Terms
Second, when there are several possible levels of relevant headings available in LCSH, it is the principle of specific entry alone that makes the choice of which level to use predictable. Works on blue crabs, for instance, could conceivably be cataloged under Crabs, Crustacea, or Chesapeake Bay, or even under Ecology, Estuaries, Invertebrates, Marine biology, Marine invertebrates, Coastal fauna, Oceanography, Arthropoda, or any of two dozen other terms—all of which appear as valid general headings in the LCSH list. The problem is that when you look in the direction of generality there is no logical or predictable stopping point, because all of the general headings could potentially apply. You could never tell which one would be the best to use. The solution is that when you search in the direction of specificity, there is indeed a predictable stopping point: the heading that is the tightest fit for what you want. The right term to stop at in the LCSH list, here, is Blue crab.
It is the predictability of the rule that creates the “control” of the controlled vocabulary—without the specificity convention, users could not know in advance which level of term to choose, and the problem of guesswork among variant headings would remain, in spite of the existence of the LCSH list. Without this rule, use of the list itself would be rendered no more efficient than having to guess which nonstandardized keywords to use, outside the list.
This, by the way, is the main problem with user-created tags: they can be, and are, assigned by different people at all levels of generality for the same book, and they are usually at very general levels. Different taggers may well assign all of the terms above to the same book. There’s no “uniform heading” or “scope-match specificity” or “specific entry” among tags.
Loss of Relationship Networks in Faceted Catalogs
A similar problem shows up when LCSH subject strings are broken up into their individual “facet” component terms. Such breakups sever the links between English language and England and Dialects and London and put the different elements in separate “silos.” Faceted catalogs—all too prevalent among research libraries these days—can be compared to disassembled watches: it’s like all of the cogs and gears have been taken apart and put separately into different bins, accompanied by the naïve advice that “now you can combine them in any way you want!” The problem is that the watch no longer works to show, systematically rather than haphazardly, what the library has when the defined relationships of its structural parts are disregarded.
Although there is an unfortunate tendency in library catalogs nowadays to lose the sharpness of their subject categorizations (due to the widespread acceptance of inadequately reviewed copy cataloging from the OCLC system, many of whose contributors do not abide by the three cataloging principles), there is still more than enough substance to the principle of specific entry that you should make use of it: as a rule, look in the direction of cross-references leading to specific headings, and stop only at the level of terminology that provides the tightest fit for your topic, rather than the general levels above it. (Look also for headings in browse menus with specific subdivisions.) Look for general levels only after you’ve first tried to be as specific as possible, rather than vice versa. If you match your retrieval technique to the rules the catalogers are supposed to follow, your results will usually be much more on target.
The principle of specific entry is essential not just as a rule for determining which headings are best for any particular topic but also for mapping the relationships of those headings to other, related topics off to the side. Again, the provision of an overview of the network of relationships among headings is entirely lost in faceted OPACs that cannot show either cross-references or browse displays of precoordinated subjects with subdivisions in a single list.
Five Ways to Find the Right Subject Headings
So, then, how exactly do you find the right heading(s) for your topic? There are five techniques: two involve using the LCSH list of subject headings, and the other three involve the use of the library’s catalog directly.
There is a “good news” and “bad news” situation here right at the start. The bad news is that many academic libraries have recently adopted OPAC search systems, mirroring Google’s single search box, that hide—or even entirely eliminate—most of the features (cross-references, browse menus, alphabetically adjacent displays, discussed below) that will enable you to get from the keywords you think up to the standardized LCSH category strings of terms. They have done this on the basis of a kind of mania sweeping the library profession these days proposing that word clouds and tags and algorithmic relevance ranking and faceted searching can replace the search mechanisms I am about to describe. They cannot, for reasons that will become apparent.
The good news, however, is that no matter where you are, you can still search the OPAC of the Library of Congress and then use the results of your subject searches there to do specific title searches for the same books in your local catalog. This freely accessible catalog is the best one available, and it approximates a model search system that should be copied more widely. The URL for the Library of Congress online catalog is catalog.loc.gov.
The following discussion will make more sense if you have an Internet connection. The “landing page” of catalog.loc.gov is where you want to be.
The five ways to get from the words you think of to the category terms used by library catalogs are the following:
1. Follow cross-references in the Library of Congress Subject Headings list, especially the NT (Narrower Term) references.
The LCSH list is an annual roster of the terms approved for use as subject headings in library catalogs. The list also includes words or phrases that are not used, with cross-references from them to the proper terms. Thus if you look up “Morality” you will find a note to use Ethics. Similarly, a search for “Surgical diagnosis,” which is not used, will refer you to the acceptable form, Diagnosis, surgical. The URL for the online list is www.loc.gov/aba/publications/FreeLCSH/freelcsh.html. This can be typed directly in your browser; an easier link to it appears on the catalog.loc.gov “landing page.” Look for the visual icon of two red books and click on “LC Subject Headings.” If you are inside a library you may also have access to the slowly dating “red books” print format copy of the list.
Once you find the proper term, the LCSH list will also show you a roster of other subject headings that are related to it so that you can systematically search either slightly different aspects of the topic or different levels of generality. Thus “Death penalty,” which is not used, refers you to Capital punishment, and under this term you will find a list of other headings that are preceded by different code designations. These codes are very important. They are UF, BT, RT, and NT (see Figure 2.1).
UF means Used For; thus, in Figure 2.1, Capital punishment in boldface type is used for “Death penalty” or “Death sentence.” In other words, if terms are preceded by UF, do not use them. They are not the acceptable search terms; instead, use the boldface heading above them. (The same printing conventions of bold and not bold are used in both the PDF and red books versions.)
BT means Broader Term(s); these are valid headings that you can search under (Criminal law, Punishment). These terms are not printed in boldface here, where they appear as cross-references, but they will be in boldface where they appear as headings in their own proper alphabetical places in the LCSH list.
RT means Related Term(s). RT references are also valid headings (Executions and executioners).
Figure 2.1 Capital punishment subject heading from LCSH red books.
NT means Narrower Term(s); these, too, are valid headings (Death row, Discrimination in capital punishment, Electrocution, etc.).
Sometimes the designation SA (See Also) will appear; it can alert you to other appearances of the heading in still other contexts.
Frequently the cross-reference USE or See will appear to get you from the wrong term(s) to the right terms, as in “Death penalty USE Capital punishment” (in the PDF LCSH list); in the online catalog itself, “Death penalty See: Capital punishment.”
There are three crucial points here, none of which is intuitively obvious. The first is that the BTs, RTs, and NTs are not subsets or subdivisions of the boldface term above. They are not included in the coverage of the boldface term; if you want any of these subjects, you must look for them directly. Thus Death row and Hanging are not included in the coverage of the term Capital punishment; if you want any of these topics you must search for them individually and directly.
The second point is that cross-references tell you explicitly which terms are “included” in the proper heading, in that all of the UF terms are effectively folded into the coverage of the boldface term under which they appear. That’s why you shouldn’t use them—they, unlike the other cross-references, are included in the uniform heading. The other cross-references (BT, RT, and NT) tell you explicitly what is not covered by the boldface heading. This is the kind of important information that semantic webs cannot provide; with them, you can never be sure of what linkages or inclusions among terms are being made—or not made.
The third point is that the NT cross-references are by far the most important ones to pursue. They are usually the specific entry terms that you need to start with. And they may lead to other, even more specific terms. Thus, within the LCSH list, Divorce does not provide a direct NT reference to Children of divorced parents, but it does start a series that leads to it. Specifically, Divorce provides an RT reference to Divorced people; this heading, in turn, provides an NT reference to Divorced parents; and this heading, in turn, provides NT references to both Children of divorced parents and Adult children of divorced parents. (The former term provides a further RT reference to the heading Children of single parents, which may be of comparative interest, once you recognize that it’s there. Cross-references, like browse menus, bring to your attention what you don’t know how to ask for until you see it.)
Knowledge of the narrower/broader nature of the cross-reference structure can help you to refine or expand your search, sometimes through an extended scale of headings, such as the following:
Descending order
Chordata
NT Vertebrates
Vertebrates
NT Mammals
Mammals
NT Primates
Primates
NT Monkeys
Monkeys
NT Baboons
Baboons
NT Hamadryas baboon
Ascending order
Hamadryas baboon
BT Baboons
Baboons
BT Monkeys
Monkeys
BT Primates
Primates
BT Mammals
Mammals
BT Vertebrates
Vertebrates
BT Chordata
Note that broader and narrower labels are always relative to other terms. For example, while Dreams, on the face of it, is a rather general term (certainly in relation to Children’s dreams), it is nevertheless a narrower term itself in relation to Subconsciousness or Visions. BT and NT designations are thus not absolute labels. A heading that is BT or broader in relation to narrower terms below can simultaneously be an NT or narrower heading in relation to broader headings above it. No matter where you enter the hierarchical sequence, however, just remember to move in the direction of the tightest-fit headings for whatever topic you ultimately have in mind.
This, then, is the first way to find the right subject headings: look in the LCSH list (either PDF or “red books”) and use the cross-reference codes, paying particular attention to the NT (Narrower Term) cross-references.
Figure 2.2 Initial search screen of catalog.loc.gov with arrow pointing to “Browse.”
Beyond the LCSH PDF display of cross-references, you can also use the LC OPAC itself (catalog.loc.gov) to show you the necessary Narrower Term cross-reference displays. To use the LCSH list within the catalog you need to click on the “Browse” option that appears immediately to the right of the blank search box (see Figure 2.2).
This will produce a new screen with a drop-down arrow next to the default “Titles beginning with” line. Click on that arrow for the two options:
The “beginning with” option is a “left-anchored” search mode; this means that whatever terms you type in will be read in the order you type them, left to right. The advantage of this option is that you need to type in only the first word in any subject-heading string in order to see all of its subdivisions. For example, if you enter “Capital punishment” alone you will immediately be given a list of associated Narrower Terms (e.g., Death row, Hanging and all subdivisions of the heading).
The “containing” option will find all of the terms you have typed in, within any subject strings, no matter what the order in which you have typed them. I’ll return to this below.
Either of these SUBJECT search options will automatically show you any Narrower Term cross-references attached to whatever terms you’ve asked for.
Both versions will also provide “See” cross-references from terms that are not used to the ones that are acceptable.
A significant difference between the OPAC (catalog.loc.gov) links to the cross-references and the PDF version of the LCSH list (www.loc.gov/aba/publications/FreeLCSH/freelcsh.html) is that the OPAC displays will not show you Broader Terms or Related Terms, whereas the PDF version will. Thus, in the OPAC you could track the descending order from Chordata (above) but not the ascending order from Hamadryas baboon. Still, in most cases the Narrower Terms are the ones most needed for effective searches.
The online catalog at the Library of Congress is a local variation of the Voyager Integrated Library System produced by ExLibris. (www.exlibrisgroup.com). The LC variation is preferable to most other online catalogs because the others have often eliminated all of the “See” cross-references that get you from the keywords you think of to the better LCSH headings and have also eliminated the browse displays that are needed to show both the additional NT cross-references and the menus of subdivided headings (as with English language—Dialects, above; or Yugoslavia, below). Without these cross-references and browse menus, the LCSH system does not work.
Two examples will illustrate their importance. (You should be online at catalog.loc.gov to follow this.) First, if you click on “Browse,” select either of the SUBJECTS options, and then type in “death penalty”—which is not the approved LCSH heading—you will immediately get a cross-reference, “See: Capital punishment.” Following that link will not only show you the many more records under the proper category term, but will also provide you with an array of cross references to even more Narrower Terms (Crucifixion, Death row, Discrimination in capital punishment, etc.) and an extensive browse menu of hundreds of narrower subdivisions of the initial topic (e.g., Capital punishment—Biblical teaching, —Bibliography, —[scores of country subdivisions], —Encyclopedias, —History, —Moral and ethical aspects, —Religious aspects, —Public opinion, etc.). The many country subdivisions are not recorded in the separate LCSH PDF list (or in the red books printed version).
Equally important, this browse menu will show you all of these subdivisions in a single list, enabling you to recognize all of the narrower options relevant to your interest in one visual roster, without endless clicking back and forth among entirely separate facet lists. (The full list will be easier to look through quickly if you change the “Records per page” box on the Browse screen from the default 25 to 100.)
In other words, in a faceted catalog display you would not be able to see the full string Capital punishment—United States—History—20th century—Case studies with its several terms all linked together; a faceted catalog separates the geographic, topical, chronological, and form subdivisions into separate silos, requiring multiple clicks if you wish to view each type, with still further complications if you wish to combine them. This is what I mean when I say that such catalogs “disassemble the watch mechanism”: it doesn’t do you any good if you can “combine the elements in any way you want” if they are no longer meshed and geared with each other. You will never be able to combine them as skillfully as the watchmaker (read: professional cataloger) assembled them, especially when facetization has destroyed the entire cross-reference network. Or again, you might consider a single-list browse menu as an overview map—you don’t want to have to look at four separate maps when all the overlapping data can be displayed on one.
In summary: the nonfacetization of browse menus in LC’s online catalog is one of the major features that make it so outstanding; the other is the inclusion of the cross-references themselves, including especially the crucial See and Narrower Term references. That’s why you may want to start with it, even before looking at your local catalog.
A second example from the LC OPAC: if you choose the initial “LC Online Catalog Quick Search”—the default “blank search box” option—and type the name of the late Libyan dictator Moammar Gaddafi as “Gaddafi” (the spelling used by, among other sources, the Washington Post), you will get only a very few records. However, if you choose Browse and then either of the SUBJECTS options and type in the same name form, you will get a cross-reference to “See: Qaddafi, Muammar.” The subsequent display will lead you not only to almost 200 records but also to a browse menu spelling out various narrower aspects of the books written on this individual (e.g., Qaddafi, Muammar—Assassination attempts, —Interviews, —Political and social views, —Religion, —Views on women, etc.) A faceted catalog will not give you the crucial cross-reference to begin with, or the browse menu—so if you have typed in the wrong term to start with, all you can do is “progressively limit” that wrong initial set, rather than be directed to the right set.
2. Look for narrower terms that are alphabetically adjacent to your starting-point term in the LCSH list. Not all narrower headings in the LCSH list receive explicit NT cross-references pointing to them. In Figure 2.3, for example, note that the NT references under the general term African Americans do not include any other headings that start with the phrase “African American(s)”; and yet there are more than a dozen pages of narrower terms that are alphabetically adjacent to the general heading.
Figure 2.3 Display from LCSH red books of terms alphabetically adjacent to “African Americans” from “African American wood carving travelers” to NTs under African American.
Thus, preceding the general heading African Americans are entries such as the following:
Other alphabetically adjacent narrower terms follow the general heading, among them:
As with the formal NT cross-references, none of these alphabetically adjacent narrower terms is “included” in the general heading African Americans; in the library catalog, each must be searched directly. In the current PDF edition of Library of Congress Subject Headings there are 13 pages of phrases starting with African American(s).
Narrower term entries do not receive formal NT linkages when they are already alphabetically adjacent to the more general terms to which they would otherwise be NT-linked. As a practical matter for researchers, this means you have two places to look for the more specific terms; the cross-reference structure alone is insufficient to alert you to all of them.
It is possible to see the full list of alphabetically adjacent terms in an online OPAC browse display; but sometimes it’s just easier and faster to look at the PDF LCSH list. On a PDF page your eye can scan laterally and diagonally, not just vertically, and can skim three columns of terms at once. Further, the displays in this list will not be cluttered with hundreds of geographic subdivisions—all of which show up in the online browse displays within the OPAC itself.
3. Within the online library catalog itself, look for “subject tracings” on relevant records that are retrieved by keyword, title, or author searches.
Sometimes a good starting point can be secured in an online catalog simply by finding one good title. To return to a previous example, a researcher looking for information on the Cockney dialect started by doing a simple keyword search for “Cockney”; this led to a list of titles that included Cockney Dialect and Slang, amid many other irrelevant (mainly fiction) hits (see Figure 2.4).
Figure 2.4 Online catalog display of record for Cockney Dialect and Slang, with arrows pointing to title and to its subject headings.
In looking at the display of the catalog record for the one relevant title he’d found, he could then discover on it the subject headings that the catalogers had added; in this case, again, the best heading was English language—Dialects—England—London. Clicking on in the proper category term then led him to dozens of other relevant titles, most of which lack the keyword “Cockney.”
This, then, is the third way to find the right subject headings for your topic: find any good record at all, by keyword, title, or author searching, and then look at what is called the subject tracing field on the display of the catalog record. (Note that in some online catalogs you may have to deliberately click on a “Full” display of the record; if the default display is “Brief” you won’t see the subject tracings.)
One advantage of using such tracings is that they will often, in effect, supply cross-references among terms that are not captured by the formal UF, BT, RT, and NT designations and which also escape alphabetical adjacency displays. For example, a book with the title Crime and the Occult: How ESP and Parapsychology Help Detection has two subject tracings:
These headings are not cross-reference-linked to each other, nor are they alphabetically adjacent, but their appearance together on the same catalog record effectively alerts researchers to the existence of both.
The main advantage of using tracings is that, in effect, doing so allows you to use any keywords you can think of as cross-references to the formal subject headings. For example, I once helped a grad student writing a dissertation on “the social construction of sexual perversions.” The LCSH system uses the unheard-of term Paraphilias in order to avoid loading the subject heading with negative connotations of perversion or deviancy, but few people if left to their own devices would ever think of this word. Nevertheless, finding a few titles of books having the keywords “sexual perversion(s)” quickly led to this standardized subject heading in their tracings. In other words, if you know how the system works, it doesn’t much matter which terms the catalogers use—“Paraphilias” was not even in the first dictionary I consulted to make sure it meant what I thought it meant—because you can still get from whatever you think of to the right headings, even when the UF, BT, RT, and NT cross-references are not adequate by themselves, through the use of tracings.
There is also a caveat here, however. Keep in mind that using subject tracings alone to find the best headings is not a good habit, although many academics fall into this rut. The problem is that the keyword searches most people start with are usually too broad. If, for example, a student interested in the effects of divorce on children simply types in the keyword “Divorce,” then the tracings on all of the retrieved records will themselves be at the wrong level of generality. Similarly, if a researcher interested in blue crabs simply types in “Crabs,” he will get hundreds of records whose subject tracings will themselves be at the wrong level of generality. They will all be valid LCSH terms—but they will also be the wrong terms, at the wrong levels of specificity, for the desired topic. It is thus bad practice to rely exclusively on subject tracings to find the right headings—if the keywords with which you start your search are not themselves accurate enough, then the tracings they lead to will also be skewed.
4. Within the online library catalog, use the left-anchored browse displays of subdivisions under a topic (via the “SUBJECTS beginning with” option in the LC OPAC using the Browse link) and be sure to look at the full array.
The fourth way to find the right subject term(s) for your topic is to look through the arrays of subdivisions that show up under appropriate headings in the catalog. This has been touched on above in the Capital punishment and Qaddafi examples, but needs more unpacking. Again, I will use the LC online catalog at catalog.loc.gov as the model of what a good catalog should do.
For example, one researcher interested in the history of Yugoslavia asked for help at the reference desk because, on his own, he’d simply done a Boolean combination of the keywords “Yugoslavia” and “history” and had been overwhelmed with way too many irrelevant records. The solution to this problem was the use of the online catalog’s browse displays. When using the Browse option on the catalog.loc.gov search page, you can then select from its drop-down menu the line saying “SUBJECTS beginning with.” If you then type in the single term Yugoslavia, a browse display of many screens’ length will automatically appear. This list includes headings such as these:
These headings provide only a sample of the full list. The researcher, in this case, was delighted: he could immediately see that, as a historian, he had many more options relevant to his interest than he had realized. He was particularly excited by many further subdivisions under Antiquities (not all listed above); these records were missed entirely by his keyword search using “history.”
This wonderful overview mapping of options would also be missed altogether if the searcher simply types “Yugoslavia” and “history” in massive full text databases such as Google Books, hathitrust.org, or Digital Public Library of America (http://dp.la). Their Web softwares cannot display overview browse menus of subjects with subdivisions.
I cannot recommend this too strongly: use your library catalog’s browse displays (or the LC catalog’s if your local OPAC lacks this feature). When there are multiple screens of subdivisions, take the time to look through all of them. You will usually be able to spot important aspects of your topic that otherwise you would never have thought of—and then narrow your selection to only those aspects. This technique is almost tailor-made to solve the frequent problem of getting too much junk via keyword searches at excessively granular levels.
The bad news here, again, is that faceted catalogs that “de-couple the strings” into separate lists of topic, time, geographic, and form facets thereby destroy the principle of specific entry—e.g., having to combine the separate facets Violence AND Women produces a general retrieval that is very different from that found under the more specific LCSH string Violence in women. Similarly, finding Afghanistan—Defenses—History—20th Century—Sources all in one string, within a browse menu, is a lot easier than having to combine Afghanistan AND History and then having to sort through those results to find the additional topical, chronological, and form aspects that together hit the nail on the head. Most people won’t take such extra steps.
The larger a library’s holdings, the more researchers must rely on menu listings that enable them to simply recognize relevant options. There are three such menus you need to look for:
A great deal of intellectual time and effort by catalogers goes into the creation of these menus; without them, you simply have to guess which terms to use; and, as in this Yugoslavia example, no one will be able to think up beforehand even a fraction of the relevant topics that could readily be of use.
Digression: Precoordination and Postcoordination
Inexperienced students of library science will sometimes say, “Libraries could save a lot of money if we simplify cataloging by eliminating multielement strings and simply record each element separately, because the computer software can combine the separate elements (or facets) into the same results.” This is the facetization problem discussed above, but there’s more to it. While it is true that the computer could indeed combine Yugoslavia AND Antiquities as separate terms, or Yugoslavia AND Bibliography, the real problem lies with the researchers themselves—the human beings who use the computers. They cannot combine the two elements when they are separate (rather than linked in a string of terms that shows up automatically in a browse display) unless it occurs to them in advance that Antiquities and Bibliography are indeed viable options. Real people cannot do this without the help of menu displays that show them the combinations they cannot anticipate.
When catalogers have taken the trouble to create linked strings, it is known as “precoordination” of terms; when the terms appear only as separate elements on the catalog record (or in a thesaurus) and have to be combined afterward by computer manipulations, it is known as “postcoordination.” (Thus, Philosophy—History is a precoordinated string; Philosophy AND History would be a postcoordinate combination of individual subject facets.) The important point is that precoordinated strings of terms offer major advantages that cannot be matched in a postcoordinate system. One is that precoordinated terms provide greater clarity of meaning and avoid retrievals of irrelevancies. A postcoordinate combination of Philosophy AND History will bring up not just Philosophy—History but also History—Philosophy. Women in advertising is not the same as Women AND Advertising. Women in communication is different from Women—Communication, and both of these precoordinated strings avoid the many irrelevancies that would be retrieved by Women AND Communication.
Also, a precoordinate system enables you to recognize, in browse displays, multiple important aspects of a topic that you could never think up beforehand. I was once asked, “How would the ancient Greeks have transcribed animal sounds?” I found a Greek dictionary that provided the information only because I could recognize a likely heading within a browse display of subdivisions under Greek language:
It would never have occurred to me in a million years to do a (postcoordinate) combination of the separate facets Greek language AND Onomatopoeic words (or just Greek AND onomatopoeic) if left to my own guesswork, but the browse menu of precoordinated strings enabled me to spot that option when I did not know how to ask for it.
This example illustrates why it is so very important to look systematically at any and all precoordinated subdivisions that appear in the library catalog. Even if there are many screens of them, skim through them all. This insight is something that took years to crystallize for me, and I would have been a much better researcher if I’d caught on to it sooner. It is not immediately obvious that you should look at the whole roster, and so most researchers just don’t do it. But I have found through experience that doing this works so well, and in such surprising ways, and in so many instances, that the technique of looking through all of the subdivision strings of a topic is a search technique that you should make use of consciously and deliberately. The provision of these browse menus in OPAC displays is one of the major advances in library science in recent decades; it solves real problems that are otherwise simply intractable in relevance-ranking keyword search systems.
Some of the many subdivisions that are possible—and that, as a reference librarian, I’ve found most useful to look out for—are these:
Obviously not all of these subdivisions will appear under any one heading, but these are among the most important ones you should be on the lookout for. Few people can remember such a list, which makes a single, simple recognition menu so crucial.
If you have a basic familiarity with the more important options, you can then look for them more deliberately. You will be a better searcher if you “get the feel” of which subdivisions you can expect to find. They are frequently overlooked in library catalogs because researchers are not taught their importance and because browse lists may sometimes be lengthy. (In faceted catalogs they won’t even be in one list; they will be segregated into entirely different geographic, topical, chronological, and form “silos.”) In using an OPAC that does have a single list of subdivisions under a heading, however, do look through the entire list. You will be able to spot the important subdivisions, such as those above, even when you’re not specifically looking for them.
Remember, too, that no matter how much we all may want simplicity in searching (“a single search box”), the reality is that the world’s books on any substantive topic are themselves immensely variegated both in their own terminologies and in their relationships to other topics. This real complexity cannot be wished away, and any retrieval system that has to deal with a highly variegated literature must itself entail some complexity to map it out. A catalog for a small public library may not need complicated search features, but a research library does.
An analogy to airplane controls may be useful: the cockpit dials and switches for a small Cessna airplane will be very simple compared to those in a huge C-5 Galaxy transport. Forcing a C-5’s pilot to work with only a Cessna’s configuration would mean he could never fly his more sophisticated plane—he could only, at best, taxi it along the ground, which would defeat the purpose of having a heavy-lift, capacious cargo airplane to begin with. No matter how much the pilot might want a simple cockpit configuration, he would not be able to accomplish his mission with a too-simple configuration. The situation is similar with research libraries’ online catalogs: if they provide only a Google-type “single search box,” or a single box followed only by facet limitations, they will thereby eliminate both cross-references and browse menus and thereby also vitiate uniform heading, scope-match specificity, and specific entry. The entire LCSH system falls apart without precoordinated strings; when the watch is disassembled into its individual component parts, it no longer works. OPACs without the complexity of browse-menu–displayed precoordinated strings and cross-references can no longer do the “heavy lifting” that scholarship requires. They cannot show you in any systematic way “what the library has” in its cataloged collection—they will give you at best only fragments of the total holdings—“something quickly” rather than “the shape of the elephant.” And it doesn’t really matter if surveys say that students want simple Google-type search boxes—the surveys don’t stick around to notice how very dissatisfied the researchers are with the results they actually get, nor do they record how grateful the students are when librarians show them the much better alternatives.
5. Use the “SUBJECTS containing” Browse option to create menus of any LCSH subject headings containing the words you want to search, no matter what the order of the words within the headings.
The browse displays of headings with subdivisions, just discussed in section 4, appear as a result of “left-anchored” searches—that is, if you simply type in African Americans or Greek language or Yugoslavia and nothing else into the “SUBJECTS beginning with” search box, the OPAC will automatically bring to your attention through a browse menu all of their subdividing aspects that you didn’t know how to ask for. If you get the first word or words in the string correct, in other words—the left-most term(s)—then the computer display will simply show you whatever comes to the right of that word (or words) without your having to type in the full length of any of the precoordinated strings.
A second way to create browse menus is also available and, for some complex inquiries, is much more useful. In the online LC catalog (catalog.loc.gov) it is accessible through the Browse option and its drop-down menu line “SUBJECTS containing.” Choosing this avenue of access allows you to type in individual words, in any order, appearing within any LCSH headings. In other words, you don’t need to get the first word in the string right, which you need to do for the left-anchored searches.
For example, not all of the books on Afghanistan are catalogued with LCSH headings that place Afghanistan at the beginning of the string. There are hundreds of other valid headings in which the name of the country is itself a geographic subdivision of other topics, whose subject strings have entirely different and unpredictable initial terms, such as:
***
***
If you have only a left-anchored “SUBJECTS beginning with” search capability ( point 4 above), you cannot call up browse menus that will show you all of the other strings relevant to the topic Afghanistan that start with a different first word at the left. “SUBJECTS containing” solves this problem.
For another example, I once helped a graduate student who was doing a dissertation on “how the several Arab–Israeli conflicts over the years have been portrayed in fiction.” Using the “SUBJECTS containing” option I could type in:
Israel? AND (Arab? OR Palestin?) AND (Fiction or Literature) NOT Juvenile
(Capital letters are not necessary; the “?” is the symbol for truncation in the LC OPAC. Thus “Israel?” retrieves not just “Israel” but also “Israeli” or “Israelis.” I added the specification “NOT Juvenile” because the students did not want to consider books written for children.) The result was an extremely useful browse display, including such headings as the following:
These various headings are not linked to each other by cross-references. Nor are they alphabetically adjacent to each other. Nor would any single left-anchored browse display bring them all up, unless you knew in advance the first word in each different string (Arab or Israel or Israeli or Judeo or Palestinian).
One more example: a researcher interested in “terrorism in India” was overwhelmed with hits in Google Web (over 3 million), Google Books (126,000), and Google Scholar (1,060), and he wanted to know if there was a better way to get an overview. For the book literature, the problem was solved by a “SUBJECTS containing” search for the combination:
India AND Terroris?
This search produced a browse menu having (among many other headings) the following:
This “SUBJECT contains” search option is thus capable of showing you parts of “the elephant”—Indian or otherwise—of the book literature on a topic that remain concealed to the four other ways of finding the right LCSH subject headings. In terms of its practical utility for researchers, this too is one of the major advances in library science in recent years. (It frequently gets overlooked, however, because it requires a library’s OPAC search system with capabilities different from those of any Web search engine.)
Miscellaneous Tips on Subject Headings
A problem that students often have at the beginning of their projects is that of narrowing down their topic to a manageable size. It should now be apparent that are several formal mechanisms in the cataloging system that enable you to do precisely this, and in a systematic manner. They are the same five ways that lead you to the proper specific (rather than general) subject headings for your topic.
A kind of “sixth way” to find the right headings is through computer combinations of headings—that is, sometimes there is no single term (or string) that expresses the subject you want, but you can still hit the nail on the head, at a specific scope-match level, by combining two or more separate headings (e.g., Mexican Americans AND Education, bilingual). Such Boolean combinations will be discussed in Chapter 10.
Here are a few miscellaneous additional tips about using Library of Congress Subject Headings:
The subject heading—or “controlled vocabulary”—searches discussed in this chapter are only one of several methods of gaining access to information. (Another related method of vocabulary control is provided by descriptors; these are like subject headings, but are used mainly in databases indexing journal articles. They will be discussed in Chapter 4.) Searches using Library of Congress Subject Headings will not work all the time for all subjects; they provide one way of gaining access to some relevant literature, primarily books. They will enable you to see much of “the shape” of the literature for that kind of material within a given library’s collection—but not necessarily, and certainly not completely, journal articles, doctoral dissertations, manuscripts, websites, or many other nonbook formats. For the sake of providing an overview of that even larger “elephant’s” parts, the ones that lie in blind spots to LCSH searches, let me anticipate a number of points to be discussed in subsequent chapters and mention here that eight alternative methods of searching can be used when no subject heading exists, or when you wish to turn up sources in addition to those found via controlled vocabulary searches:
Each of these approaches—like that of searching with controlled vocabulary terms—has its own advantages and disadvantages. Collectively viewed, each has strengths that compensate for weaknesses in the other search techniques. An awareness of this basic structure of options, with only these few distinct methods of searching, can have a profound effect in substantially increasing the efficiency of your research, no matter what subject area you are working in—and regardless of whether you have any prior subject expertise in that area.