A word is not the same thing as one of its meanings
How can people translate Wikidata Lexeme glosses quickly while preserving the correct meaning? Begin with the lexeme’s language, lemma, lexical category, forms and distinct senses, then translate the explanatory gloss attached to the exact sense ID rather than blindly replacing a word with a familiar equivalent. Wikidata lexical data describes words and their meanings in a structured form. Its lexeme entries are not ordinary multilingual labels on encyclopaedia items, and they are not a promise that one word in language A equals one word in language B in every context.
Consider the English word ‘bank’. It can refer to a financial institution or the side of a river. If a contributor adds a target-language gloss for the financial sense under the river-bank sense ID, the target text may be fluent and grammatically correct but the structured dictionary becomes wrong. Learners retrieving the data may be taught an invalid meaning relationship. Automated tools may then propagate the confusion into a vocabulary exercise, a search index or a machine-assisted translation suggestion. The error is not a spelling defect; it is a sense-selection defect.
The safe fast workflow is to inspect the lexeme and its existing sense list, identify the concept using context and evidence, write a concise target-language gloss that defines rather than merely substitutes, and preserve the lexeme and sense identifiers during review. Searches such as ‘translate Wikidata lexeme gloss’, ‘Wikidata senses versus forms’, ‘lexeme translation statements’ and ‘Wikidata multilingual dictionary data’ describe related but distinct steps in this work. This article focuses on one reader job: adding accurate explanatory language to an existing sense. It explains when new lexemes, sense-to-sense links or form data are different tasks and why they must not be smuggled into gloss translation.
Understand Wikidata’s three different kinds of identity
Wikidata has item entities, property entities and lexeme entities, among others. An item typically refers to a concept, person, location or thing and can have labels in many languages. A lexeme refers to a lexical element of a particular language, such as a word, affix or expression. The lexeme has its own identifier beginning with L, and its forms and senses have identifiers derived from that lexeme. This structure lets the project represent words as linguistic objects rather than simply as translations of item labels.
A lexeme identifier might resemble L12345; a form identifier follows a pattern like L12345-F1, and a sense like L12345-S1. These strings are identifiers assigned to particular entities, not words a translator is supposed to rewrite. In public examples, real identifiers must be checked rather than assumed to refer to the intended word. It is safer to use symbolic example IDs in teaching notes and inspect a live lexeme before editing. An impressive target-language gloss attached to the wrong -S identifier remains a data error.
The lexeme has a language and a lexical category. A noun and a verb with identical written forms can be distinct lexical entries, and languages may organise categories differently. The lemma is a dictionary-style written representation, not a universal cross-language title. Forms represent grammatical or written realizations of the lexeme, while senses represent meanings. Glosses describe those meanings in one or more languages. Translation QA becomes much easier once editors stop treating these components as interchangeable text fields.
A lemma is not a space for arbitrary translated words
The lemma is usually the dictionary headword or another conventional form by which speakers recognise the lexical entry. A Wikidata lexeme can have multiple lemmata in appropriate spelling variants or scripts, but that is not the same as attaching foreign translations to an entry as if they were alternate spellings. The source lexeme’s language remains part of its identity. If an English word has a French semantic counterpart, the French word normally belongs to a French lexeme, not as a random extra English lemma on the original entry.
This distinction matters for scripts and dialects. A lexeme may have several representations in accepted orthographies or writing systems within its modelling conventions. Those representations are not necessarily different meanings. Conversely, two expressions that look similar may be different words or senses. Do not infer a lexeme’s identity solely from a shared sequence of letters. A translator should consult the language-specific Wikidata modelling guidance and lexical sources before deciding whether a spelling variant belongs to one lexeme, a separate lexeme or a form entry.
For a gloss-localisation assignment, leave lemmata unchanged unless the user or community has explicitly asked for an orthographic correction. The translator’s target language belongs in the sense’s gloss field, where the explanation can be expressed naturally. A gloss in Malay explaining an English lexeme does not turn the English lexeme into a Malay word. It makes that English sense understandable to a Malay reader. That is a powerful contribution, but it works precisely because the entry’s linguistic identity is preserved.
Forms express grammar, not a list of synonyms
Forms represent inflections or other grammatical realizations of a lexeme. An English noun may have singular and plural forms, while a verb may have tense or participle forms. Other languages can have grammatical gender, case, number, agreement, classifiers, aspect or far richer morphological patterns. Wikidata form representations are attached to form identifiers and can have grammatical features represented by linked items. The data model is meant to describe the language, not to force every language into English-style boxes.
When editing a sense gloss, avoid changing forms to make the translated explanation sound smoother. A source verb’s plural or tense form is not the translated target-language verb. The English form ‘runs’ and the French form ‘court’ are not two representations of one English form simply because they can occur in similar sentences. They belong to the grammatical systems of their respective languages. Treating forms as a bilingual word list corrupts morphology and can make downstream conjugation or vocabulary applications return nonsense.
Some glosses need grammatical explanation, particularly when the sense is restricted to an idiom, informal register or specialised domain. That explanation belongs in the definition and relevant supporting data, not as an arbitrary feature change. A translator may note that a sense occurs only in a certain usage, but should not invent formal inflections unsupported by evidence. If the underlying lexeme data lacks a needed form, raise a separate modelling issue. The clean boundary between meaning explanation and grammatical form maintenance is essential for accurate structured language resources.
Senses are the centre of the gloss-translation job
A lexeme can have several senses, and each sense has a stable ID. A gloss describes the meaning of that particular sense using natural language. For a polysemous word, a separate gloss for each sense helps readers choose among meanings. A Wikidata sense can have glosses in multiple languages without becoming a different sense. The task is to add another explanation of the same meaning, not to merge adjacent senses or to promote a rough equivalent into an identity claim.
Begin with the source sense’s existing glosses and any relevant statements. Where the gloss is very short, inspect example usage or reliable dictionaries and ask a knowledgeable speaker which interpretation is intended. Words such as ‘charge’, ‘current’, ‘case’ and ‘light’ can refer to several unrelated concepts. A gloss that simply repeats the lemma often fails to disambiguate them. A good target-language gloss should enable a reader to recognise which sense would fit a sentence and which similar sense would not.
Suppose the English source gloss means ‘the flow of electric charge’ for a sense of ‘current’. A translation saying only ‘flow’ is too broad. A translation describing the speed of moving water is a different domain. The correct gloss should preserve the physical concept and distinguish it from a water current or a current fashion trend. A bilingual subject specialist may be needed because the terminology is scientific. The point is not to maximise explanation length, but to provide the minimum wording that makes this sense identifiable to the intended reader.
Test the sense using a context sentence
An efficient sense-selection technique is to create a short source-language sentence demonstrating the intended meaning, then ask whether the target gloss would help a learner understand that sentence. For ‘bank’ in a river context, the sentence might involve walking along the water’s edge. A target gloss describing an institution that accepts deposits fails immediately. For ‘bank’ in a financial context, a sentence about opening an account selects the other sense. These contexts need not be published as claims about the word’s entire usage; they serve as disambiguation tests during translation.
Context matters even for apparently simple words. ‘School’ may refer to an institution, a building, a group of fish or a philosophical tradition. ‘Class’ could refer to a lesson group, an object-oriented programming construct or a social category. If the target gloss does not distinguish the intended sense, an unfamiliar reader may choose the most common meaning in their language and misunderstand the entry. A concise explanation can include the functional difference without becoming an encyclopaedia article. The reviewer should ask whether someone who did not know the English word could select the appropriate use.
Add a negative test: would the translated gloss also accidentally fit a neighbouring sense? If so, refine it. For example, a target gloss meaning simply ‘a place to learn’ might describe both a school and a university, while the intended lexeme may refer specifically to a type of school. A definition need not list every exclusion, but it should identify enough characteristics to avoid collapsing important categories. This method turns subjective translation review into a practical comprehension task with inspectable evidence.
Distinguish translating a gloss from linking translations
Wikidata supports sense-related statements, including a property used for translations between lexical senses. Such a statement links one specific sense to another sense, which is a stronger structured claim than adding a gloss in another language. The existence of a similar word in a dictionary does not prove that it matches every nuance of the source sense. If the assignment is simply to explain an existing sense to a new audience, add or review the target-language gloss and leave cross-lexeme links unchanged unless there is a separate evidence-based request to update them.
A sense-to-sense translation relation requires a target lexeme and an appropriate target sense. Selecting the target word by its lemma alone is unsafe when that word is polysemous. A target language may divide a source concept into several narrower expressions or use one broader word for multiple source senses. Register, regional usage, connotation and part of speech can further complicate equivalence. An editor should not create a machine-readable link merely because a translation service suggests a familiar word. The precise source and target senses must be compared with evidence.
This distinction is valuable to children learning vocabulary as well as contributors editing structured data. A gloss tells us what a word means in a particular use. A translation relation suggests which word sense in another language is an appropriate counterpart. These are connected intellectual tasks but have different proof requirements. Confusing them promotes the false belief that dictionaries are tables of interchangeable words. The lexeme model instead helps us represent how language carries distinctions that do not always line up neatly across communities.
Language tags do not replace language knowledge
Lexeme lemmata and form representations carry language-related spelling-variant information; sense glosses are language-tagged explanations. The specific supported codes and editing interfaces must be checked against Wikidata’s documentation. A broad language tag can be useful, but a specialised script or orthographic variant may require a more precise representation when the project’s modelling conventions support it. Choose codes to describe the actual language of the gloss, not the user’s country or the language of the source lexeme.
A gloss written in Malay about an English sense is a Malay gloss of an English lexeme. Its language tag does not mean that the source lexeme changes language. This sounds obvious but catches a common confusion when editors use a multilingual interface: the UI language, the lexeme language and the gloss language are separate variables. Someone browsing Wikidata in French may be editing an English lexeme and adding a Chinese gloss. All three dimensions can differ, and each should remain explicit in the review notes.
Do not assume every transliteration is a translation. A phonetic rendering of an English proper name in another script may be a spelling or naming convention, not an explanation of a lexical sense. Likewise, a target-language gloss that repeats the source word in another script may not help readers distinguish its meaning. Decide whether the work is orthographic representation, definition, translation relation or user-interface localisation. A job becomes faster when the field to edit corresponds exactly to the question being answered.
Write glosses as definitions, not as marketing descriptions
A good sense gloss explains a lexical meaning concisely. It should not add promotional adjectives, examples of usefulness or personal advice that are absent from the concept. For an education term such as ‘tutor’, a gloss might describe a person who provides instruction or academic support, subject to the sense’s scope. It should not claim that the tutor guarantees examination success or that the word refers only to a particular school level if the source sense is broader. Those additions turn a definition into a claim about practice or performance.
Aim for clarity without stripping away a necessary distinction. A gloss meaning ‘a person who teaches an individual or a small group’ may be more accurate for one sense than merely ‘teacher’. In another language, one everyday term may cover tutor, coach and mentor. When that happens, add a short descriptive phrase to convey the boundary. The measure of success is whether the reader can identify the correct sense, not how closely the gloss imitates the source word order.
Avoid circularity. A gloss of ‘interpret’ that says ‘to interpret’ gives no information to someone who does not know the word. The same problem occurs with overly broad descriptions such as ‘a thing used in science’ or ‘an action involving money’. Identify what sets the sense apart, ideally using simpler familiar language. If specialised terminology is unavoidable, a target-language subject expert can verify that it means the same thing and is appropriate for the educational audience. The editor should preserve a brief evidence note so later contributors understand the definition choice.
Worked example: energy, power and a tempting shortcut
A science vocabulary project contains separate lexical senses for ‘energy’ and ‘power’. Students sometimes treat the two words as interchangeable. A translator reviewing one sense finds a target-language everyday word commonly used for both and adds the same vague gloss to each entry. The result seems efficient, but it damages the distinction the dictionary was meant to preserve. In physics, energy and power are related but different quantities; power describes the rate at which energy is transferred or used. A definition that makes them identical teaches the wrong concept.
The repair is to inspect each source sense and its unit or domain context. Where the intended sense is physical, translate ‘energy’ with a definition that expresses a physical capacity or quantity in context, and ‘power’ with a rate-related definition. A domain reviewer should confirm the local scientific terminology used in the relevant curriculum. The gloss need not contain an entire textbook chapter, but it must not erase the essential conceptual boundary. One practical test is to ask whether the gloss explains why a measurement of power cannot simply be substituted for a measurement of energy in a problem.
A second review considers everyday senses. ‘Energy’ can describe liveliness, and ‘power’ can describe authority or influence. Those meanings may require separate glosses from the physics senses. A translator who sees only one headword and chooses the most familiar target equivalent can contaminate all senses. Review sense IDs individually and document which conceptual context each one represents. Students gain more from understanding these distinctions than from memorising a list of target words with no explanation of when to use them.
A second failure: grammatical forms were mistaken for meanings
An editor wants to translate the English word ‘run’ into another language and opens a lexeme with several forms. They add target-language words for ‘running’ and ‘ran’ as if each were a gloss of a distinct sense. The forms relate to grammar, not necessarily different meanings. The new entries may mix tense, aspect, noun usage and unrelated senses such as operating a machine. The structured data becomes hard for both learners and machines to interpret. The error arises from reading a list of visible forms as a list of translations.
Repair it by separating the questions. Which lexeme is being described? What lexical category does it have? Which forms represent inflections of that lexeme? Which senses distinguish meanings such as moving on foot, operating equipment or managing a business? Then provide target-language glosses for those senses without altering the original form inventory. If a target-language lexeme is needed for a translation relation, find or create it under the appropriate community process with its own language, lemma, forms and senses. Do not use the English form records as a shortcut.
The transferable insight is that a language resource has multiple axes. Form answers ‘How does this lexical element appear in a grammatical context?’ Sense answers ‘What can it mean?’ Translation answers ‘How is a particular meaning expressed in another language?’ Some words share forms but not meanings; other words share meanings but have different grammatical behaviour. Accurate vocabulary work keeps these axes visible. That is what allows a structured dictionary to support more sophisticated learning than one-to-one word substitution.
Community editing is a reviewed data contribution
Wikidata is collaborative. Existing entries may have incomplete glosses, outdated wording, modelling debates or vandalism. The presence of a public edit button does not mean every plausible translation should be added immediately. Read the project’s lexicographical data guidance and relevant language community conventions. Search for existing senses before proposing new ones. When uncertain, prepare offline suggestions with links to evidence and ask experienced speakers or editors to review. A broad machine-generated batch can create significant cleanup work if its sense mapping is wrong.
Record source information for unusual meanings. A reliable dictionary, corpus attestation, published textbook or domain standard can help disambiguate a term. Do not confuse a web search snippet containing a word with evidence for a particular definition. Proper nouns, trade names, regional slang and technical jargon may have more constraints than ordinary words. If the available evidence is weak, leave an unresolved note rather than inventing a crisp but unsupported gloss. Honest uncertainty is more valuable than a false exact answer in lexical data that other applications may reuse.
Keep an audit trail that identifies the original lexeme and sense, proposed locale gloss, reviewer, evidence and date checked. When a contribution is made to the community, use its established editing process and observe the site’s licence and attribution requirements. The offline manuscript here explains how to prepare and QA such contributions; it does not authorise making live edits to Wikidata or any website. Separating research and proposal from publication is also useful for classroom projects, where students can learn to analyse meaning before deciding whether they have sufficient evidence to contribute publicly.
Query the right records, not just a familiar word
The Wikidata Query Service can retrieve lexemes, their forms and their senses using distinct predicates. The official lexicographical data query documentation explains that ordinary item-label mechanisms do not automatically retrieve lexeme lemmata or sense glosses. A user who runs a general item search for the word ‘bank’ might find an item about finance instead of the lexeme whose senses need translation. Search should therefore begin with the right entity type and a reviewed lexeme ID, then inspect its sense list. Treat search results as candidates, not proof that a particular occurrence is the intended meaning.
The RDF model uses ontolex:sense to associate a lexeme with senses and skos:definition for glosses. When querying gloss languages, the language tag on the definition can be tested. The query below is deliberately illustrative: it shows the general relation but will return rows only for lexemes already represented in the service. It does not decide whether a row is correct or whether the gloss is suitable for a particular learner. Human editors still need to compare definitions with real usage.
SELECT ?lexeme ?sense ?gloss WHERE {
?lexeme ontolex:sense ?sense .
?sense skos:definition ?gloss .
FILTER(LANG(?gloss) = "en")
}
LIMIT 20
This can help a reviewer see that senses are separate records rather than interchangeable labels. A more focused review would filter to the exact lexeme under consideration and include other gloss languages when available. Avoid using one broad query as a mass-translation instruction. The service can show data already recorded, but it cannot infer the intended sense for every word in a new sentence. A contributor should decide whether the source gloss is adequate, whether another sense already exists and what evidence supports a proposed target-language definition before preparing an edit.
A practical ledger for one language pair
Build an offline review sheet with lexeme ID, sense ID, source gloss, relevant example sentence, domain, lexical category, proposed target gloss, target language code, source reference and reviewer decision. Add a column for near-neighbour senses that must not be confused. This is especially valuable for polysemous words: it forces the reviewer to test the definition at the level of the sense record rather than the headword. A termbase can then be reused safely only when both source meaning and domain match.
One column should record whether the source gloss is ambiguous or appears incomplete. A translator is not necessarily the person authorised to clarify a defective source definition. They can propose a clearer target interpretation, but the source editor or community should decide whether the underlying sense needs revision. If two possible meanings remain, the row should not be published as fully reviewed. A temporary ‘needs disambiguation’ status is more honest than accepting whichever machine translation sounds most fluent.
The ledger also supports consistency across related lexemes. If the same grammatical or scientific concept appears in several entries, approved target terminology can be reused where the sense really is the same. Where the intended distinctions differ, record the reason for different target wording. It is easy to create a glossary that looks consistent by flattening meaning; the better goal is principled consistency that respects domain, register and lexical category. A future editor should understand why a particular gloss was accepted without having to repeat every conversation from scratch.
Review meaning with three complementary tests
The first test is positive context. Ask whether the target gloss clearly explains the lexeme’s use in a representative sentence for that sense. The second is negative context. Ask whether it wrongly fits a neighbouring sense or an unrelated homograph. The third is target-language naturalness. Ask a proficient speaker whether the gloss is a credible explanatory phrase rather than a literal calque or a marketing tagline. All three matter because a definition may sound natural while describing the wrong thing, or accurately paraphrase the source while being incomprehensible in the target language.
For a scientific or legal sense, add expert review. The word ‘force’ can refer to physical interaction, coercive power or a group of people. A bilingual physicist can distinguish its technical use from everyday usage, but a general translator may select the most familiar equivalent. Likewise, a source gloss about ‘liability’ needs precision about responsibility and obligation rather than a phrase that merely suggests inconvenience. The task of a lexeme gloss is not to issue professional advice; it is to represent a lexical meaning. However, specialist vocabulary still requires evidence and domain knowledge.
Use a small set of representative examples to test the proposed gloss, not a single carefully chosen sentence that guarantees success. If the sense applies in both a school textbook and a professional report, see whether the target explanation works in both. If the sense is restricted to slang or a regional dialect, do not broaden it to general formal usage. Reviewers should log these limits. A good definition teaches readers when a word applies and, by implication, when a different sense is needed.
Build an ethical and maintainable batch review
Batching saves time when editors group records by language, subject and evidence source, but speed should not override per-sense inspection. Automated extraction can identify missing target glosses or repeated source terms. A spreadsheet can make patterns visible. Neither operation establishes semantic equivalence. Review each proposed gloss against the source sense and a trustworthy context before publishing. Begin with a small sample so that systematic errors in word-sense disambiguation can be detected before hundreds of records are affected.
Protect community trust by distinguishing suggestions from verified contributions. An offline file may contain hypotheses prepared by a language model or translation service. Label them accordingly and route uncertain entries to human reviewers. Do not claim that a gloss has been verified merely because another online dictionary shows a similar word. Cross-check polysemy, register, regional usage and grammatical role. If two sources disagree, preserve that conflict in the review note instead of silently choosing whichever is easier to translate.
When a target language lacks a compact equivalent, a descriptive gloss may be appropriate. A cultural institution, local practice or specialised grammatical feature might require several target-language words. That is not necessarily a failure of translation; the purpose of a gloss is understanding. A literal one-word replacement could create a false claim of equivalence. Explain the concept briefly, avoid unsupported analogies and seek the language community’s preferred terminology. This is a practical way to respect linguistic diversity instead of measuring quality by word-for-word symmetry.
Changed-context transfer exercise: a technology vocabulary list
An education team is assembling a bilingual list for the words ‘cell’, ‘current’, ‘charge’ and ‘model’. One contributor finds the English lexemes and adds a single target-language word to each lemma field. Another writes the same short gloss for every sense of ‘cell’, covering biology, electricity and prison buildings. A third edits a verb form because they think its suffix should match the target translation. The students will use the resulting list to answer science questions. How should the project be corrected before release?
First restore the identity boundaries. The English lexemes keep their language and lemmata; their grammatical forms are not slots for target words. For each target gloss, select the appropriate sense ID. The sense of ‘cell’ referring to a biological unit is different from the sense referring to an electrochemical cell or a prison room. The target explanation must preserve that domain distinction. If a target-language lexeme is needed as a lexical counterpart, find and review the corresponding lexeme and sense rather than inserting the word as another source lemma. Revert unrelated form edits unless there is a separately justified linguistic correction.
Next design the learning check. Provide a sentence about a battery cell and a sentence about a biological cell. Ask students which target gloss fits each, and why. For ‘current’, ask them to distinguish the flow of charge from the flow of river water. For ‘model’, distinguish a mathematical representation from a physical miniature when those are separate intended senses. A high-quality list is one in which students can use words accurately in new situations, not simply repeat bilingual pairs. The Wikidata sense structure supports that deeper learning when editors maintain its distinctions faithfully.
Prepublication inspection for each proposed gloss
Confirm the lexeme language, lexical category and specific sense ID. Read all existing glosses for that sense where possible, review neighbouring senses and check at least one trustworthy contextual source. Ensure the proposed target gloss is tagged for the language in which it is written, accurately defines the chosen meaning, avoids circular wording and does not overstate the source. Confirm that no lemma, form representation, grammatical feature or sense-to-sense relation was changed unintentionally. If a structural edit is genuinely needed, split it into a separately reviewed proposal.
Compare near-neighbour records. A proposed gloss should not collapse distinctions between ‘apply’ and ‘approve’, ‘theory’ and ‘hypothesis’, ‘measurement’ and ‘estimate’, or ‘energy’ and ‘power’. Try positive and negative sentences. Ask a proficient target-language reviewer to judge clarity, and ask a domain expert where a specialised concept is involved. Store links or citations supporting unusual terminology, plus a note about uncertainty. If the target explanation cannot be defended, leave it in draft rather than giving it the appearance of vetted public data.
For any eventual community contribution, check the interface and current local guidelines again. An offline spreadsheet or article cannot prove that a public record has not changed since extraction. Another editor may have added a better gloss or split a sense. Reconcile against the live lexeme immediately before proposing an edit, and preserve the current identity and revision context. This reduces accidental overwrites and duplicate contributions. A structured publication gate is especially useful for projects preparing many lexical entries for multiple languages.
What this method does not establish
A translated gloss does not guarantee that two languages partition meaning in the same way. A sense-to-sense relation may be approximate, contextual or contested. A lexeme’s existing sense model may itself be incomplete or disputed by editors. It is important to explain these limits to students and dataset users who might assume that a structured database produces uniquely correct equivalents. The database provides a framework for organising linguistic knowledge; careful evidence and ongoing community judgement determine how reliable its content becomes.
Nor can a short gloss replace the full teaching of pronunciation, syntax, register, collocation, idiom and cultural context. A student may understand what ‘charge’ means in physics yet still misuse it in a sentence. Forms, usage examples and instructional practice help build that broader competence, but they are different tasks from translating the explanatory gloss of an existing Wikidata sense. Keeping the scope bounded makes it possible to deliver high-quality work instead of promising that one metadata field solves every difficulty of language learning.
Summary: translate a meaning, not a string in isolation
The reliable path through Wikidata lexical translation is lexeme first, sense second, target gloss third. Keep lemmas and forms in their own grammatical roles, preserve identity-bearing IDs and select a meaning using evidence. Make the gloss natural enough for its audience but precise enough to distinguish neighbouring senses. If a target lexical equivalent is required, review the exact corresponding sense in that language as a separate structured claim. This method helps make vocabulary a useful technology for understanding and reasoning, rather than a collection of deceptively simple word pairs.
Frequently asked questions
Is a Wikidata item label the same as a lexeme lemma?
No. An item label names an entity in the item model; a lexeme lemma is a linguistic representation of a lexical entry in a particular language.
Should I add a foreign word as another lemma of an English lexeme?
Generally no. A semantic counterpart in another language normally belongs to a separate lexeme; glosses can explain the English sense in other languages.
What is a Wikidata sense ID?
It is the identifier of one meaning within a lexeme, typically written with the lexeme ID followed by an -S component.
How are forms different from senses?
Forms describe grammatical or written realizations; senses describe meanings. One word may have several forms and several distinct senses.
Can one sense have glosses in several languages?
Yes. Language-tagged glosses allow speakers of different languages to understand the same selected sense.
Does adding a gloss make a sense-to-sense translation link?
No. A cross-lexeme translation statement is a separate structured relationship requiring comparison of particular senses.
Why is a dictionary-equivalent word sometimes insufficient?
It may cover several senses, miss a register restriction or imply a different cultural or technical concept. A gloss should explain the exact intended meaning.
Can AI-generated glosses be published without review?
That is risky. They require sense-level evidence, target-language review and compliance with current community guidelines.
How do I check a technical gloss?
Inspect domain usage, compare neighbouring senses and have a qualified bilingual reviewer assess whether the crucial concept is preserved.
What should a good offline batch record?
Lexeme and sense identifiers, proposed gloss, target-language tag, evidence, reviewer, remaining ambiguity and current revision status.
Authoritative sources
- Wikidata Lexicographical Data Documentation — lexeme identifiers, lemmata, forms, senses, glosses and interface guidance.
- Wikidata Lexicographical Data Query Service — querying lemmata, forms and language-tagged sense glosses.
- MediaWiki WikibaseLexeme RDF Mapping — representation of lexemes, forms and senses in linked data.
- Wikidata Lexicographical Data Portal — overview and related community conventions.
Related eduKateSG reading
- How People Translate Quickly | Drupal Interface Translation QA — language strings in an application, distinct from lexical senses.
- How People Translate Quickly | Directus Translations Interface QA — stable content entities and locale relationships.
- How People Translate Quickly | Moodle Language Customisation QA — interface vocabulary rather than multilingual dictionary meaning.