GLAS7999 Chap.3 Build a Search and Source Strategy
Build a Search and Source Strategy
Search in loops and select by function
The course treats searching as movement between concepts, results, bibliographies and a revised question. This chapter builds a concept matrix, uses explicit discovery and citation passes, and records why sources are kept or rejected.
It then classifies source functions as direct, theoretical, contextual, methodological or broader debate, so the bibliography serves several research decisions instead of repeating the same case study.
Translate the question into concepts
Do not paste the full research question into a catalogue and treat the result count as a verdict. Separate the object, relation, place, period and analytical frame.
Under each heading, list synonyms, historical labels, spelling variants, romanisations and terms used in adjacent fields. Combine one term from different columns, observe what changes, and delay narrow filters until the vocabulary is understood. A geographic label may hide work catalogued under empire, diaspora, language or a former place name.
The concept matrix makes those assumptions visible and shows which word is actually controlling discovery.
Use passes with different purposes
A discovery pass learns the field’s language and recurring authors. A focused pass combines that vocabulary around the project’s relation and boundary. Citation chasing moves backward to foundations and forward to later revisions or disputes.
Treat these as distinct operations rather than repeating one query with minor punctuation changes. Record the database or catalogue, exact query, date, filters and outcome. Irrelevant results can reveal that a term has another disciplinary meaning; empty results can reveal an anachronistic category.
A query log turns both outcomes into design information and prevents accidental repetition.
Assign sources jobs
Direct sources address the same or a closely related object. Theoretical sources supply concepts, contextual sources explain settings, methodological sources model an operation, and broader-field sources place the case in a larger debate.
These labels describe what a publication does in this project, not its permanent identity. One article may serve two roles, but each role should be named. A bibliography of five near-identical case studies creates apparent depth while leaving theory, context or method untested.
Ask what decision the next source must help make, then search for that function rather than for another topical match.
Make rejection reproducible
Keep a short reason for every serious exclusion: wrong period, incompatible object, inaccessible language, duplicated evidence, review rather than research, or method unable to support the required inference. Avoid the empty label not useful.
State what job the source might have performed and why it cannot perform it here. This record stops a famous or convenient publication from returning merely because it is familiar.
It also distinguishes principled scope from accidental omission and allows a later revision of the question to reactivate sources whose role has changed.
Follow networks without inheriting them
Reference lists reveal earlier conversations and citation indexes reveal later responses, but citation networks reproduce a field’s exclusions as well as its canon.
When the same authors recur, test alternative subject headings, languages, journals and regional catalogues. Look at who is cited only for data, who is credited with theory and whose work appears outside the dominant database. The aim is not symbolic variety detached from relevance.
It is to discover whether the apparent shape of the debate is an artefact of the search route and to ensure that a selected bibliography can challenge its own starting assumptions.
Stop with an explicit debt list
Searching cannot become literally complete.
Stop when new combinations mostly repeat known works, major source functions are covered, and each important design choice has scholarship capable of supporting or contesting it. Then write what remains unresolved: a missing primary source, untranslated debate, untested method, absent counter-position or inaccessible archive. These are search debts, not reasons to conceal uncertainty.
Some will guide the next session; others will become stated limitations. A stopping rule makes sufficiency arguable and keeps bibliography building subordinate to the research problem.
What this chapter covers
- 01
Concept matrix
- 02
Multilingual search terms
- 03
Database passes
- 04
Citation chaining
- 05
Keep or reject record
- 06
Five source functions
Worked example · free
Balance a selected bibliography
- 2Identify the imbalance.
- 3Name missing functions.
- 3Justify the next searches.
Key terms
- Concept Matrix
- A table of core concepts, synonyms, translations and variants used to build queries.
- Citation Chaining
- Following references backward and later citations forward from a strong source.
- Source Function
- The job a source performs in a particular project.
Build a Search and Source Strategy FAQ
How many search terms should I use at once?
Combine one term from different concept columns and vary them systematically. Putting every synonym, place and period into one query can produce no results without showing which concept blocked discovery.
When can I stop searching?
Stop when new searches mainly repeat known work, major source functions are covered and each important design choice has literature capable of supporting or challenging it. Record remaining search debts as explicit limits.
Can one source have several roles?
Yes, but name the role that matters for this project. A case study may provide direct evidence and a methodological model, yet those uses should be distinguished so later citations support the right claim.
Assessment move
Maintain a search log with query, database, useful vocabulary, keep or reject decision and source role. At the end of each session, revise one question term and write the next unresolved design choice before searching again.