This chapter teaches students how to move beyond simple keyword searching into the more powerful territory of controlled vocabulary, subject headings, database filters, and citation chaining. Students will learn why keywords alone often fail to capture the full scope of a topic, how subject heading systems organize scholarly knowledge into stable, searchable categories, how database filters and facets help narrow results without changing search terms, and how citation chaining — both backward into a source's references and forward into sources that cite it — builds research networks that keyword searching alone cannot construct.

By the end of this chapter, students should be able to use subject headings strategically, apply filters and facets with purpose, trace citations in both directions, and build a source network that reflects the actual scholarly conversation on their topic rather than a random sample of keyword matches.

Before you can understand why subject headings and citation chaining are necessary, you need to understand exactly where keyword searching breaks down. Keyword searching fails in several predictable ways, and recognizing those failure modes is the first step toward correcting them.

Academic subjects have multiple names. Legal scholars, sociologists, economists, and public health researchers may all study the same population but use different disciplinary vocabularies to describe them. A historian writing about child labor in nineteenth-century England may use terms like "factory children," "climbing boys," or "pauper apprentices" — period-specific language that does not match the modern keyword "child labor" that a beginning researcher would type. A psychologist writing about depression may use "major depressive disorder," "unipolar depression," "clinical depression," or simply "MDD" depending on the decade and the journal. A keyword search built around one of these phrases will miss the others.

The vocabulary problem is not limited to historical or technical language. It exists wherever scholars disagree about how to name something, where field-specific jargon has replaced everyday language, or where a topic sits at the intersection of multiple disciplines that use different terminology. Environmental science, public health, legal studies, and economics all study climate change, but their vocabularies differ in ways that make cross-disciplinary keyword searching unreliable.

Keyword searches also suffer from a precision problem: they retrieve everything that contains your search words, regardless of whether those words are the main topic of the source. A keyword search for "water" in a history database might retrieve articles about irrigation, naval warfare, public health infrastructure, the symbolism of water in literature, the chemistry of water treatment, and the theology of baptism. The word "water" appears in all of them, but most are not about water as a research topic in any unified sense.

Adding more keywords helps but does not fully solve the precision problem. "Water rights" is more specific than "water," but it still retrieves sources where "water rights" appears anywhere — in a footnote, in a passing reference, in a title that mentions water rights only as context for another argument. Subject headings, by contrast, are assigned to sources where the topic is central, not incidental.

Keyword searching is also incomplete in a way that is invisible to the searcher. You cannot know what you missed, because you can only see what your search returned. A keyword search for "prison education" might return forty results. But if thirty other relevant sources are indexed under "correctional education," "incarcerated students," "prison literacy programs," or "education in detention facilities," you will never see them unless you happen to search those exact phrases too. The forty results you found may look like a complete picture. They are not.

Subject headings address the completeness problem by gathering all sources on a topic under one consistent term, regardless of the words individual authors used. If you find the right subject heading, you find everything indexed under it — not just the subset that happens to match your particular word choices.

There is one more limit of keyword searching that is harder to name: it provides no browsing structure. When you search by keyword, you get a ranked list of matches. You cannot easily see the shape of a subject, the subcategories within it, the adjacent topics that might be relevant, or the broader contexts that frame it. Subject headings, by contrast, are organized into hierarchies — broader terms, narrower terms, and related terms — that give you a map of a subject rather than a list of keyword hits. Browsing that map often surfaces relevant topics and approaches that you would never have thought to search for directly.

Controlled vocabulary is a system of standardized terms used to describe and retrieve information consistently. Unlike natural language, which varies across speakers, writers, time periods, and disciplines, controlled vocabulary is fixed: one term is chosen to represent each concept, and that term is applied consistently across an entire indexing system.

The alternative to controlled vocabulary is natural language indexing, which is what most internet search engines use. A search engine like Google retrieves pages that contain the words you typed, regardless of whether those pages are actually about your topic. The engine has no controlled vocabulary; it just matches strings of text. This works well for many everyday searches but poorly for systematic academic research.

Controlled vocabulary systems work by making a single decision about terminology and enforcing it. If the controlled vocabulary for a database decides that the correct term for research on people without legal immigration status is "undocumented immigrants," then every article in that database about that population will be indexed under that term — whether the article's own author used "undocumented," "unauthorized," "illegal," or "undocumented noncitizen." As a searcher, you need only know the controlled term to find all the relevant articles.

Controlled vocabulary is not a flat list of unrelated terms. It is a structured system with internal relationships. Most controlled vocabulary systems include three types of relationships:

These relationships are the architecture of controlled vocabulary. They let you navigate a subject up, down, and sideways — broadening your search if you are not finding enough, narrowing it if you are finding too much, and exploring adjacent areas if your initial angle is not productive.

Different databases and library systems use different controlled vocabularies. Some of the most common include:

Each database uses its own controlled vocabulary, which means the terminology that works perfectly in one database may not work in another. A researcher who learns to use MeSH in PubMed will need to learn different terminology for searching in JSTOR or a sociology database. This is one reason why starting a search in a new database always involves some initial exploration — you need to find the controlled vocabulary the database uses before you can search with real precision.

Subject headings are the form of controlled vocabulary most relevant to library catalog and database searching. A subject heading is a standardized phrase assigned to a source by a professional indexer or cataloger to describe the source's primary topic or topics.

Subject headings are not tags or keywords that authors apply to their own work. They are assigned by trained people following consistent standards, which means they are more reliable and more precise than the author's own choice of words. A book titled A River in Winter might be entirely about salmon habitat loss in Pacific Northwest streams, but you would never find it through a keyword search for "salmon" or "habitat conservation." If the cataloger assigned the subject heading "Salmon — Habitat — Pacific Northwest," you can find it instantly.

Subject headings in the Library of Congress system follow a specific structure: a main heading followed by optional subdivisions that specify the heading further. Subdivisions are separated by dashes.

These subdivisions can be combined in various orders to produce highly specific headings. A heading like Women — Employment — United States — History — 20th century is far more precise than the single word "women" would be as a keyword, and it retrieves only sources indexed under exactly that combination.

When you open a catalog record for a book or an article record in a database, subject headings appear in a designated field, usually labeled "Subject," "Subject Headings," or "Descriptors." In library catalogs using LCSH, these headings appear in the full record and are often hyperlinked so that clicking on them runs an immediate search for other items with the same heading.

The key habit is this: when you find one relevant item, always open its full record and read all of its subject headings. Then click on the most relevant heading to retrieve all other items in the catalog or database assigned that heading. This single step often produces better results than a dozen additional keyword searches.

Subject heading subdivisions also serve as precision tools. If "Poverty" returns thousands of results and you need material specifically about poverty and health in sub-Saharan Africa, searching the heading "Poverty — Health aspects — Africa, Sub-Saharan" is far more efficient than adding keywords like "health" and "Africa" to a general search. The subdivision structure is built by catalogers who have already done the organizational work for you. Using it saves you from sorting through irrelevant results manually.

Once you have a set of search results — whether from a keyword search, a subject heading search, or some combination — database filters and facets let you refine those results without submitting an entirely new search. Filters are one of the most practical tools in academic database searching, yet many students scroll past them without using them at all.

A filter is a limiting option that narrows a set of search results to only those that meet a specific criterion. Common filter types include:

A facet is similar to a filter but works slightly differently in interface design. Facets typically appear as sidebar panels showing categories and counts — for example, "By Publication Date: 2020 (143), 2019 (97), 2018 (88)..." — and you narrow results by clicking on a category. Facets are dynamic: clicking one facet narrows the results and updates all other facets to reflect the new, smaller set. Filters and facets accomplish the same function; the terms are sometimes used interchangeably depending on the database.

The strategic use of filters depends on your research needs at a given stage of the process.

The peer review filter deserves special attention because it is widely used and widely misunderstood.

Marking a source as "peer reviewed" in a database filter does not guarantee that the source is high quality, accurate, or free from error. Peer review is a process, not a quality certification. It means that the manuscript was sent to other experts in the field for anonymous evaluation before publication, and that those reviewers recommended acceptance (with or without revisions). But peer review is imperfect: reviewers can miss errors, hold biases, fail to replicate studies, or simply approve work that is methodologically weaker than ideal.

Peer review does mean that the source went through more structured expert scrutiny than a blog post, a news article, or a self-published book. In academic research, peer review is a reasonable baseline requirement for scholarly sources. But the peer review filter should be understood as a starting point for evaluation, not a substitute for it.

The most powerful use of filters involves combining them.

Different databases offer different filter options, reflecting their subject focus and collection design. A biomedical database like PubMed offers filters for clinical trial type, age group, sex, and species — criteria important for medical research but irrelevant in a literature database. A law database may filter by jurisdiction, court level, or type of legal document. A newspaper database may filter by section, author, or city edition.

When you enter a new database, take a few minutes to explore what filters it offers before you begin searching. Understanding the available filters helps you plan your search strategy more efficiently.

Citation chaining is one of the most powerful and underused research strategies available to academic researchers. It works by following the citations within and around a source rather than conducting additional keyword searches. There are two directions: backward (into a source's past) and forward (into a source's future). This section covers backward chaining; the next covers forward chaining.

Every academic book, article, and report contains citations — references to earlier works that the author built on, responded to, quoted, disputed, or synthesized. These citations are not merely footnotes. They are a map of the scholarly conversation that preceded the source you are reading. Backward citation chaining means following that map.

When you read the bibliography or reference list of a source and identify earlier works that appear relevant to your research question, you are beginning a backward citation chain. You are looking upstream from your source — finding the works that led to it.

Backward chaining works for a fundamental reason: scholars build on each other. A 2022 article about housing policy and health outcomes did not emerge from nothing. It builds on decades of prior research — studies from the 1970s establishing the relationship between housing quality and respiratory health, studies from the 1990s documenting health disparities in public housing, studies from the 2000s examining natural experiments when housing policies changed. All of that foundational research appears in the 2022 article's citations.

If you only find the 2022 article through keyword searching, you have one source. But if you follow its citations, you have access to the entire lineage of scholarship that made that article possible. The bibliography of a well-researched scholarly source is a curated reading list on your topic, assembled by a scholar who has already done the work of identifying which prior sources are most important.

Not every item in a bibliography is worth following. Backward chaining requires selectivity. Scan a bibliography for: works whose titles directly relate to your research question; works published by well-known scholars in the field; works published in high-quality journals or by major academic publishers; works cited specifically in the sections most relevant to your question; and works described in the text as "foundational," "influential," "seminal," or "definitive."

You do not need to trace every citation. You need to identify which citations are most worth your time, retrieve those, and then continue the chain selectively.

Backward chaining is especially valuable for historical research and for understanding how a field's thinking on a topic has evolved. By chaining backward from a recent article to a decade-old study to a classic work from the 1960s, you can trace the intellectual history of an idea — seeing which arguments came first, which were challenged, which were revised, and which became foundational. This historical perspective enriches your understanding of the topic far beyond what any single current source can provide.

It also helps you avoid a common research error: presenting as new something that is actually long-established, or ignoring foundational debates that still shape current scholarship. Backward chaining surfaces the historical context that keyword searches of recent literature often miss.

If backward chaining follows sources into their past, forward citation chaining follows sources into their future. It answers a different question: not "what did this source build on?" but "who has cited this source since it was published?"

When a scholarly work is published, other scholars read it, respond to it, apply it, challenge it, extend it, and correct it. All of that subsequent engagement appears in the reference lists of the later works. Forward citation chaining uses tools that track these citations to find sources that have built on your original source since it was published.

This is especially powerful for older or foundational sources. A study published in 1990 may have been cited by hundreds of more recent works. Following those "cited by" links takes you directly to the most current scholarship on the topic, written by scholars who found the original work important enough to engage with.

Forward chaining is particularly valuable in three situations: when your source is several years old and the field may have moved substantially since then; when you are trying to understand impact, since citation volume offers a sense of how a source was received by the scholarly community; and when keyword searching has saturated, since forward chaining from your best sources often produces new, relevant material that no keyword search would have found.

Backward chaining retrieves older sources. Forward chaining retrieves newer ones. Together, they create a temporal arc — a chronological narrative of how a topic has developed in the scholarly literature from its origins to the present. This combination is especially powerful for research papers that require demonstrating familiarity with the full scope of a scholarly conversation, not just a snapshot of recent keyword matches.

Beyond citation chaining through individual sources, researchers can use bibliographies and related-source tools to navigate the scholarly landscape more efficiently. These tools work differently from citation chaining but serve complementary purposes.

A standalone bibliography is a compiled list of sources on a topic, assembled not as part of a book or article but as a reference work in its own right. Bibliographies are valuable because they represent expert curation: a scholar or librarian has already done the work of identifying relevant sources, filtering out less relevant ones, and often annotating entries with descriptions of content and value.

Many libraries, including the MGU Library, publish subject guides — also called research guides, LibGuides, or pathfinders — that compile the most useful databases, key journals, foundational texts, and reference sources for a specific discipline or topic. These guides are assembled by subject librarians who specialize in that field and update them regularly as the field evolves.

A well-constructed subject guide is not just a list of databases. It is a research network: a structured map of the most important resources in a field, organized to help researchers navigate from background knowledge through primary research into the deepest specialist literature. Using a subject guide before you begin searching saves you from discovering the most important resources one at a time through trial and error.

Many academic databases include "Related Articles," "Similar Items," or "More Like This" features that identify other database records related to a source you are viewing. These features use algorithms to identify sources that share subject headings, keywords, authors, or citation patterns with the source you are looking at.

Academic texts — especially in humanities disciplines — use footnotes not just to cite sources but to conduct scholarly conversation. A footnote may credit an idea, dispute a claim, acknowledge a complicating counterargument, direct the reader to a deeper treatment of a subtopic, or note important works the author chose not to engage with at length. Reading footnotes carefully is a form of citation chaining that operates at the argumentative level, not just the bibliographic one.

When a footnote says something like "For a contrasting view, see [Author]" or "This argument draws on [Author]'s influential account of..." that is the author telling you exactly where the most relevant intellectual connections are. Following those footnotes takes you to the heart of the scholarly conversation, not just its surface.

The strategies in this chapter — subject headings, filters, backward citation chaining, forward citation chaining, bibliographies, and related records — are most powerful when used together, as a coordinated approach to building a source network rather than a collection of unrelated search techniques.

A source network is the connected set of sources that address your research question from multiple angles, time periods, disciplinary perspectives, and evidentiary types. It is not a random accumulation of keyword matches. It is a structured understanding of who has written what about your topic, how those sources relate to each other, where they agree, where they disagree, and where the most important debates lie.

Building a source network takes more time than compiling a keyword search results list. But it produces something qualitatively different: a researcher who has built a source network understands the field. A researcher who has only compiled keyword matches has a pile of documents.

How do you know when your source network is mature enough to support your research? Several signals indicate that you have reached a productive level of coverage:

Reaching this point does not mean you have read everything. It means you have built a network comprehensive enough to support responsible, well-grounded research. That is the goal.

The reason source networks matter goes beyond research efficiency. A source network is the foundation of synthesis — the process of combining sources to form a broader understanding that addresses your research question. You cannot synthesize sources you have not found. You cannot identify agreements, disagreements, patterns, and gaps across a literature you have only sampled at the surface through keyword matching.

The MGU Synthesis Model — which you will study in depth in Part VI — depends on having a source network that is broad enough to reveal agreements, disagreements, patterns, and gaps. You need sources from multiple perspectives to compare perspectives. You need sources from multiple time periods to trace development. You need sources addressing different aspects of your topic to construct a complete picture.

Choose a research topic from the list below or use a topic from a current research project: the psychological effects of social media use on adolescents; land rights and Indigenous communities in North America; the history of antibiotic resistance; urban heat islands and public health; or criminal justice reform and recidivism.

The following is a sample database record for a peer-reviewed article. Read it carefully and answer the evaluation questions.

Use the following model when recording a subject heading and citation chaining session in your research notebook.

Use citation counts as one data point among several, not as a substitute for evaluating a source's methods, evidence, and reasoning. The MGU Source Evaluation Model — Relevance, Authority, Evidence, Accuracy, Purpose, Currency, Context — is the right framework for evaluation. Citation counts may inform your assessment of authority and impact, but they do not replace it.

When you find sources through subject heading searching, citation chaining, or related source tools, record them in your research notes immediately. The retrieval path matters as much as the source itself.