A Research Stack for Korean Learners: Corpora, Dictionaries, White Papers, Archives

Core examples: 말뭉치; 사전; 백서; 통계자료; 법령; 표준; 학술논문; 기록관; 공문서; 보도자료; 출처 기록; 검색어; 메모.

The problem: search results are not a research method

A learner meets a difficult Korean term in news, a government document, and a forum post. Search results disagree. One blog gives a casual explanation. A dictionary gives a broad meaning. A government page uses the term in a specific policy context. A forum uses it sarcastically. Which one should the learner trust?

The answer is not “trust the first result.” Serious Korean research requires a stack of resource types, each with a job.

Resource categories

Dictionaries define headwords, part of speech, pronunciation, examples, Hanja, and standard forms. Corpora show usage patterns, collocations, genre distribution, and real examples. Government white papers and reports define institutional concepts and policy terms. Legal texts define obligation, permission, prohibition, and eligibility. Standards define technical terminology and procedure. Academic articles define research terms and argument traditions. Archives preserve older usage, records, newspapers, and historical context. Official forms show practical labels and required actions. Media databases show public framing.

No single resource answers every question.

What each resource is good for

Use a dictionary for initial meaning and form. Use a corpus for usage and collocation. Use official documents for institutional meaning. Use legal texts for legal force. Use academic papers for specialized analysis. Use news for public discourse and framing. Use archives for history. Use forms for action language.

For example, the word 자격 can be defined in a dictionary, found in corpus examples, specified in an eligibility notice, formalized in a legal text, and debated in news. Each source adds a layer.

Research discipline

Record sources. Record date accessed. Record genre. Record authority level. Record uncertainty. Mark whether the source is prescriptive, descriptive, official, academic, commercial, journalistic, or personal.

A Korean learner’s research note should say: “This term appears in housing notices with X meaning; official source uses it in Y context; news uses it more broadly; learner action: recognize in documents, do not use casually yet.” That is better than copying one English gloss.

Levels of stack

Intermediate learners need dictionaries, learner dictionaries, simple corpora, and official notices. Advanced learners add monolingual dictionaries, news archives, white papers, legal texts, and academic abstracts. Translators add standards, parallel texts, terminology databases, and domain experts. Country researchers add statistics, government reports, archives, and media discourse.

Technical-review guardrail: authority depends on the question

An official source may define a policy term but not everyday usage. A corpus may show common usage but not legal correctness. A news article may show public framing but not technical definition. Match source authority to the question.

Remediation upgrade: resource stacks need authority labels

The v2 pass gives the research-stack article a stricter evidence hierarchy. Dictionaries, corpora, white papers, statutes, standards, academic papers, archives, official forms, press releases, and forum posts do different work. A forum can show current informal usage, but not institutional definition; a statute can define legal force, but not everyday tone.

The article now tells learners to record authority level, date, genre, source owner, descriptive versus prescriptive function, and confidence. This prevents random search results from outranking better evidence simply because they are easier to read.

Mini practice: choose the resource

Learner workflow: research stack note

1. Define the research question.
2. Search dictionary for base meaning.
3. Search corpus for usage patterns.
4. Check domain-specific official sources.
5. Add news or academic sources if public framing or theory matters.
6. Record source type, date, and confidence.
7. Write learner action: memorize, recognize, verify, ask expert, or defer.

Suggested functions:

1. Question classifier: meaning, usage, legal force, historical, domain, public discourse.
2. Resource recommendation: dictionary, corpus, official, legal, academic, archive.
3. Source log: URL/title, date, genre, authority, note.
4. Confidence meter: low, medium, high, needs expert.
5. Term note export: usable glossary entry.

Final rule

Do not let random search results become your Korean research method. Match the resource to the question, record the source type, and treat uncertainty as data.