Creating Domain-Specific Korean Glossaries From Source Texts

Core examples: 용어집; 전문용어; 출처; 정의; 문서 장르; 연어; 신청; 자격; 제출; 계약; 진단; 개인정보; 위험도; 검증.

The problem: ordinary word lists collapse in real domains

A learner enters Korean real estate, medical, immigration, technology, insurance, or academic texts and discovers that general vocabulary is not enough. Words like 신청, 자격, 제출, 계약, 진단, 개인정보, 동의, 해지, 갱신, 보증금, 체류자격, 증빙, 유의사항 do not behave like isolated dictionary entries. They belong to document systems.

A domain glossary is not just a word list. It is a map of terms, actions, documents, collocations, risks, and source evidence.

Glossary fields

A strong entry includes term, domain, definition, source sentence, document genre, collocations, Hanja or loanword note if useful, risk level, translation note, and verification status.

For example:

• 제출서류: immigration/municipal/school forms; documents required for submission; collocates with 확인, 준비, 누락, 보완, 첨부; high action relevance.
• 계약갱신: housing/work/contracts; renewal of contract; collocates with 요구권, 기간, 만료, 임대차; legal risk.
• 선택 동의: app/privacy; optional consent; collocates with 철회, 마케팅 정보 수신, 접근권한; privacy relevance.

Extracting terms from source texts

Start with five source documents from the same domain. Highlight recurring nouns, headings, form labels, warning phrases, verbs that control action, and terms connected to eligibility or obligation. Do not collect every unknown word. Collect terms that decide what someone must do, may do, must submit, must pay, may receive, or must avoid.

In a municipal notice, 대상, 신청기간, 신청방법, 제출서류, 선정, 발표, 문의처 are core. In a rental listing, 보증금, 월세, 관리비, 옵션, 입주 가능일, 전입신고, 확정일자 matter. In app privacy, 수집·이용, 제3자 제공, 선택 동의, 철회, 접근권한 matter.

Quality control

Verify terms across multiple sources. Separate everyday meaning from domain meaning. 신청 can be ordinary “application,” but in a form it means a procedural act with deadline and documentation. 자격 can mean eligibility, qualification, or status depending on domain.

Mark uncertainty. A glossary that admits uncertainty is more useful than one that guesses confidently. Use labels: verified, likely, needs review, high-risk.

Document function matters

The same term changes force by document type. 계약 in an advertisement, contract, legal explainer, and complaint form has different weight. 개인정보 in a privacy consent screen differs from 개인정보 처리방침. 진단 in a hospital context differs from a business “diagnosis” report.

Record document genre so the word does not float free from its function.

Technical-review guardrail: domain glossaries are not professional advice

A learner glossary helps reading and question formation. It does not replace lawyers, doctors, immigration officers, safety professionals, accountants, or official agencies. High-risk terms should be labeled as high-risk and verified through authoritative sources.

Remediation upgrade: domain glossaries are literacy tools, not expert credentials

The v2 pass clarifies that a domain-specific glossary helps the learner read documents; it does not qualify the learner to make legal, medical, housing, immigration, finance, safety, or technical decisions alone. Risk level and source type should be fields in the glossary, not afterthoughts.

The article now recommends separating everyday meaning from domain meaning, storing the full source sentence, recording document genre, and marking uncertainty. A good glossary tells the learner when to continue reading and when to seek official or professional confirmation.

Mini practice: build the entry

Learner workflow: 50-term domain glossary

1. Collect five real source documents from one domain.
2. Highlight repeated headings, labels, warnings, and action verbs.
3. Select 50 terms that control action or understanding.
4. Add source sentence and document genre.
5. Verify with dictionaries, official materials, or multiple documents.
6. Mark risk level and uncertainty.
7. Review terms by document scenario, not alphabetical list only.

Suggested functions:

1. Term extraction fields: term, domain, genre, source sentence.
2. Collocation collector: verbs, nouns, particles, formulae.
3. Risk label: low, medium, high, official action required.
4. Verification tracker: dictionary, corpus, official source, expert review.
5. Scenario review: read a new document using saved glossary.

Final rule

A Korean domain glossary should preserve context. Do not collect naked words; collect terms with source, genre, collocation, action relevance, risk level, and verification status.