{
  "id": "lexicography",
  "title": "Lexicography",
  "lang": "en",
  "source_lang": "en",
  "summary": "Lexicography is the practice and study of compiling dictionaries, encompassing lexical evidence, entry design, definitions, translation equivalents, and digital reference systems.",
  "category": "language",
  "subcategory": "",
  "type": "concept",
  "aliases": [
    "Dictionary-making",
    "dictionary compilation"
  ],
  "generated_at": "2026-11-04",
  "model": "pro/gpt-6.1-sol",
  "url": "https://aiwiki.page/en/lexicography/",
  "translations": {
    "zh-Hans": {
      "url": "https://aiwiki.page/zh/lexicography/",
      "json": "https://aiwiki.page/zh/lexicography.json"
    }
  },
  "sections": [
    {
      "heading": "",
      "text": "Lexicography is the practice of designing, compiling, editing, and maintaining dictionaries, together with the systematic study of their contents, structures, production, and use. Practical lexicography produces reference works; theoretical lexicography, often called metalexicography, investigates how such works represent vocabulary and serve their users. Its objects include not only printed wordbooks but also historical dictionaries, bilingual resources, specialized glossaries, learner’s dictionaries, and continuously updated digital services.\n\nLexicography draws on linguistics, especially the study of vocabulary and meaning, but it is not simply the application of a linguistic theory. A dictionary must transform incomplete and variable evidence into information that readers can find, interpret, and use. Decisions about which expressions to include, how to divide their meanings, what examples to provide, and how to label their usage combine empirical analysis with editorial judgment.\n\nDictionaries support reading, writing, translation, education, and research, while also influencing public perceptions of linguistic correctness and cultural belonging. Their development connects ancient lexical lists with historical scholarship, large electronic corpora, and computational language technologies. Across these settings, a central problem remains how to represent a changing vocabulary accurately within the constraints of a particular reference work."
    },
    {
      "heading": "Scope and intellectual foundations",
      "text": "Lexicography concerns lexical units rather than merely strings of letters separated by spaces. Its subjects include simple words, compounds, affixes, idioms, fixed expressions, abbreviations, and conventionalized combinations. The distinction between a written word and a lexeme is fundamental: write, writes, wrote, and written can be treated as forms of one lexical unit, whereas the noun record and the verb record may require separate grammatical descriptions. Expressions such as by and large demonstrate why dictionary coverage cannot be restricted to single orthographic words.\n\nThe discipline is closely related to lexicology, the study of vocabulary as a linguistic system. Lexicology investigates such matters as lexical organization, word formation, and relations between meanings; lexicography must decide how to present those matters in a usable reference resource. The relationship is reciprocal. Dictionaries supply organized evidence for linguistic research, while linguistic analysis helps editors distinguish senses, recognize phraseological patterns, and describe grammatical behavior. Nevertheless, a theoretically elegant account is not necessarily the clearest explanation for a schoolchild or language learner.\n\nSeveral complementary approaches inform dictionary research. Linguistic approaches emphasize the adequacy of descriptions of form, meaning, and use. User-oriented approaches examine the situations in which consultation occurs: understanding an unfamiliar expression, choosing a word while writing, checking pronunciation, or identifying a technical concept. Functional approaches distinguish dictionaries according to the assistance they provide in communicative or knowledge-related situations. These perspectives explain why there can be no single ideal entry for every audience.\n\nLexicography also overlaps with terminology, bibliography, information science, and publishing. Terminological work commonly begins with concepts in a specialized field and identifies their designations; general lexicography more often begins with expressions and investigates their uses. This contrast is useful but not absolute. Technical dictionaries may combine concept systems with descriptions of ordinary linguistic variation, while general dictionaries routinely cover scientific and professional vocabulary.\n\nThe boundary between a dictionary and an encyclopedia is likewise permeable. A dictionary conventionally explains linguistic expressions, whereas an encyclopedia describes their referents or subjects. Yet defining kangaroo requires information about an animal, and an encyclopedic account must identify the terms it employs. The practical distinction lies in selection, emphasis, and access structure rather than in a complete separation between linguistic and world knowledge. Digital reference services can combine both kinds of information while maintaining separate editorial responsibilities."
    },
    {
      "heading": "Historical traditions and major transformations",
      "text": "The earliest surviving lexical documents belong to the scribal cultures of ancient Mesopotamia. Lists written in cuneiform organized signs, occupations, objects, and other categories for education and administration. Later bilingual lists connected Sumerian expressions with Akkadian equivalents. These works were not modern general dictionaries, but they established enduring practices: selecting lexical material, imposing an order, and using one linguistic form to explain another.\n\nChinese lexicography developed several complementary organizational traditions. The Erya, a composite work with ancient origins, groups expressions under semantic headings. The Shuowen Jiezi, compiled by Xu Shen around 100 CE and presented to the imperial court in 121, analyzed characters and arranged them under 540 graphic headings. Later dictionaries developed radical-based and phonological arrangements. Such traditions show that alphabetical order is only one solution to lexical access: the properties of a writing system can shape the organization of an entire dictionary.\n\nSouth Asian lexical scholarship included synonymic and thematic vocabularies, notably the Sanskrit Amarakośa, commonly dated to approximately the fifth century CE. Arabic lexicography developed sophisticated arrangements related to consonantal roots and phonological structure. The Kitāb al-ʿAyn, associated with al-Khalīl ibn Aḥmad in the eighth century, organized material through an ordering of sounds rather than the familiar written alphabet. These traditions addressed literary interpretation, religious scholarship, language teaching, and the preservation of learned vocabulary.\n\nIn medieval Europe, glosses explaining difficult Latin words developed into independent glossaries and bilingual collections. The expansion of printing made larger and more standardized reference works commercially feasible. Ambrogio Calepino’s Latin dictionary appeared in 1502 and was repeatedly expanded in multilingual editions. Institutional projects included the first dictionary of the Accademia della Crusca in 1612 and the first dictionary of the Académie française in 1694. Robert Cawdrey’s A Table Alphabeticall of 1604 explained difficult English words rather than attempting comprehensive coverage of ordinary vocabulary.\n\nSamuel Johnson’s Dictionary of the English Language appeared in 1755, combining definitions with extensive literary quotation. Noah Webster’s American Dictionary of the English Language followed in 1828. During the nineteenth century, historical dictionary projects increasingly sought to document vocabulary through dated evidence. The first fascicle of what became the Oxford English Dictionary appeared in 1884; its first edition was completed in 1928, and a supplement followed in 1933. The brothers Grimm’s Deutsches Wörterbuch began publication in 1854, and its first edition was completed in 1961.\n\nTwentieth-century innovations included dictionaries designed specifically for foreign-language learners and the systematic use of electronic corpora. A. S. Hornby’s learner-oriented dictionary, first published in Japan in 1942, helped establish an influential model emphasizing meaning, grammar, and practical usage. The COBUILD dictionary, first published in 1987 under John Sinclair’s direction, made corpus evidence central to its descriptions. Subsequent online publication altered the unit of production from a periodically completed volume to an incrementally revised lexical database."
    },
    {
      "heading": "Dictionary types, audiences, and purposes",
      "text": "Dictionaries can be classified along several independent dimensions. A work may be monolingual or multilingual, general or specialized, contemporary or historical, comprehensive in aspiration or deliberately selective. These categories overlap: a bilingual dictionary may specialize in law, while a monolingual dictionary may describe only regional speech. Format is another dimension rather than a dictionary type in itself. The same editorial database can support a large website, a compact mobile interface, and a selective printed edition.\n\nGeneral-purpose dictionaries typically provide spelling, pronunciation, grammatical classification, meanings, and usage information for a broad vocabulary. Historical dictionaries organize evidence diachronically, tracing changes in form and meaning through dated citations. Etymological dictionaries concentrate on origins and relationships between words. A thesaurus supports access through meanings or semantic associations rather than primarily through an unfamiliar word form. Pronouncing dictionaries, spelling dictionaries, and reverse dictionaries isolate particular consultation needs.\n\nSpecialized dictionaries describe vocabulary associated with fields such as chemistry, law, music, or computing. Their intended users strongly affect the required detail. A technical definition suitable for specialists may rely on concepts inaccessible to beginners. A dictionary for translators may need equivalents, contextual restrictions, and institutional distinctions that are unnecessary in a glossary written for practitioners within one jurisdiction. Specialized lexicography therefore requires both subject expertise and attention to how terms function linguistically.\n\nLearner’s dictionaries are designed for people acquiring an additional language. They often use a controlled defining vocabulary, explicit grammatical patterns, frequent examples, and information about collocation. A native speaker may already know that commit a crime is conventional and do a crime is generally less idiomatic; a learner may require that contrast to be made visible. Dictionaries for children face different constraints, including reading ability, curricular relevance, and the need to explain common words that adult dictionaries may assume are already understood.\n\nDictionary use is often divided into receptive and productive tasks. Receptive consultation supports understanding something encountered in speech or writing. Productive consultation supports constructing an expression, selecting an equivalent, or checking acceptability. A brief synonym may solve a receptive problem while failing to show how a word combines with complements or which social settings permit its use. Consequently, the amount of information in an entry is less important than its suitability for the intended task.\n\nUser research studies these distinctions through observation, interviews, controlled tasks, questionnaires, and digital interaction records. Search logs reveal common queries but do not by themselves establish whether users understood the results. A successful consultation may involve finding a spelling, distinguishing two senses, or recognizing that no direct equivalent exists. Evaluation therefore requires evidence about the user’s problem as well as the resource’s contents."
    },
    {
      "heading": "Selection, headwords, and dictionary structure",
      "text": "A dictionary’s macrostructure is its overall inventory and arrangement of entries. Its microstructure is the organization of information within an entry. A further dimension, often called the mediostructure, consists of cross-references linking entries and other components. These distinctions separate three editorial problems: what the work contains, how each unit is described, and how users move between related pieces of information.\n\nSelecting the headword list requires a policy on coverage. Frequency is important, but a dictionary consisting only of the most frequent forms would omit many expressions that readers need help understanding. Editors also consider distribution across genres, duration of use, regional significance, technical importance, and the likelihood of consultation. Proper names, transparent compounds, offensive expressions, and new abbreviations require explicit inclusion policies. A dictionary’s omission of a word is therefore not evidence that the word does not exist.\n\nLemmatization groups related forms under a citation form or lemma. In English, a verb is usually entered under its base form, while nouns are generally entered under a singular unless their lexical behavior favors another treatment. Languages with richer inflection require additional decisions about predictable paradigms and irregular forms. A reader encountering an irregular form may need a separate access entry even when the substantive explanation appears elsewhere.\n\nHomograph division and sense division are related but distinct. Two identically spelled items may receive separate entries because of different origins, different grammatical categories, or an editorial preference for shorter entries. Within one entry, meanings may be ordered historically, by frequency, or according to a semantic hierarchy. None of these arrangements is universally best. Historical ordering supports diachronic research, while frequency ordering may allow a learner to reach the most likely meaning faster.\n\nAlphabetical order also requires conventions. Editors must specify how to treat spaces, hyphens, numerals, diacritics, and variant spellings. Chinese dictionaries may use radicals, stroke counts, or pinyin; dictionaries of other scripts employ their own established sequences and access systems. Digital search can combine several routes, but the underlying data still require a coherent treatment of forms and variants.\n\nEntry design must balance completeness with navigability. Pronunciation, grammatical information, definitions, examples, labels, etymology, and cross-references compete for attention. Print dictionaries economize through abbreviations and compressed syntax. Digital dictionaries can reveal information progressively, but excessive nesting or unexplained icons can make an entry harder to interpret. Consistent presentation is therefore an integral part of the lexical description, not merely a matter of visual styling."
    },
    {
      "heading": "Evidence, corpora, and editorial workflow",
      "text": "Lexicographic evidence may come from published texts, recorded speech, archival manuscripts, specialist documentation, fieldwork, and reports submitted by readers. Historical projects traditionally assembled quotation slips recording a form, its context, and bibliographic details. The essential function of such a record survives in digital systems: every claim about usage should be traceable to evidence that editors can inspect and evaluate.\n\nModern corpus linguistics supplies tools for examining large collections of language. A text corpus is not simply a large quantity of text; its composition and documentation determine what conclusions can be drawn from it. A collection dominated by news reporting will represent journalistic vocabulary more fully than household conversation. A corpus of internet material may contain duplicated pages, machine-generated text, and uneven demographic coverage. Size cannot compensate automatically for these distortions.\n\nA concordance displays occurrences of a search item with surrounding context. Editors use concordance lines to identify recurrent meanings, grammatical constructions, collocations, and characteristic settings. Frequency counts indicate how often a form appears, while dispersion measures indicate whether its occurrences are spread across sources or concentrated in a small number of documents. A technical term repeated thousands of times in one manual does not have the same general distribution as a word appearing moderately often across many genres.\n\nAutomated annotation can improve retrieval. Part-of-speech tagging helps distinguish, for example, nominal and verbal uses, while lemmatization gathers inflected forms. Syntactic parsing can identify recurrent subject–verb or verb–object relations. These processes introduce errors, particularly with rare vocabulary, historical spelling, and informal writing. Editors therefore inspect original contexts rather than treating automatically generated summaries as unquestionable evidence.\n\nA typical production workflow begins with a project specification and entry-selection policy. Researchers collect evidence; lexicographers draft entries; subject specialists or language experts review difficult material; and editors check consistency, cross-references, and presentation. Large projects use style guides to regulate definitions, labels, sense numbering, citations, and grammatical notation. Versioned editorial systems preserve changes and distinguish proposed content from approved publication.\n\nHistorical evidence requires additional caution. The earliest surviving citation is the earliest example currently documented, not necessarily the moment when an expression was invented. Newly digitized archives can produce antedatings, and unreliable document metadata can create false ones. Contemporary evidence raises a parallel issue: a conspicuous new expression may be short-lived or confined to a narrow community. Editorial decisions weigh persistence and distribution rather than relying only on publicity or raw search-result counts."
    },
    {
      "heading": "Meanings, definitions, and semantic organization",
      "text": "Defining a word requires distinguishing its conventional contribution from information supplied by a particular context. Lexical semantics provides concepts for this task, including polysemy, semantic relations, and patterns of extension. Yet dictionary senses are editorial models of usage, not directly observable objects. A word such as paper can refer to a material, a newspaper, an article, or an examination document; deciding how many numbered senses to assign depends partly on the dictionary’s purpose and level of detail.\n\nEditors sometimes describe contrasting tendencies as “splitting” and “lumping.” Splitting gives distinct treatment to relatively fine differences; lumping groups related uses under broader explanations. Excessive splitting creates long entries whose distinctions users cannot reliably apply. Excessive lumping obscures differences in meaning or grammar. Corpus evidence helps by showing whether a proposed sense has recurrent contextual patterns, but frequency alone does not determine semantic boundaries.\n\nA traditional definition identifies a broader category and a distinguishing property. A triangle, for example, can be defined as a plane figure bounded by three straight sides. This genus-and-difference method works well for many categories, but it is less straightforward for function words, emotional vocabulary, discourse markers, and culturally variable concepts. An explanation of although must describe a relation between propositions rather than identify a class of physical objects.\n\nOther techniques include synonym definitions, paraphrases, usage explanations, and full-sentence definitions. A full-sentence formulation can incorporate a typical participant structure: “If someone postpones an event, they arrange for it to take place later than originally planned.” Such a definition supplies information about both meaning and use. Learner-oriented definitions must also avoid unnecessary dependence on vocabulary more difficult than the item being explained.\n\nCircularity is a persistent problem. Defining two unfamiliar terms solely through each other provides little assistance. Controlled defining vocabularies reduce this risk, although a small vocabulary cannot eliminate all dependence among concepts. Technical accuracy introduces another constraint: an accessible explanation must not erase distinctions essential to the field. Editors may combine a plain-language definition with a more precise note rather than force both functions into one sentence.\n\nSense relations and semantic change affect entry organization. Metaphor can extend a bodily expression to an abstract domain, while metonymy connects associated entities, as when a place name stands for an institution. Repeated extensions may become conventional enough to merit separate treatment. Historical dictionaries document semantic change through dated evidence; contemporary dictionaries describe the distinctions relevant to present usage. Examples help establish those distinctions by showing meanings in contexts that rule out competing interpretations."
    },
    {
      "heading": "Grammar, pronunciation, phraseology, and usage",
      "text": "A definition alone rarely provides enough information to use a word correctly. Dictionaries also describe morphology, syntax, pronunciation, spelling, and restrictions on context. A verb entry may identify transitivity, irregular past forms, permissible complements, and recurring prepositions. For example, the patterns depend on something, avoid doing something, and persuade someone to do something cannot be derived reliably from a brief synonym.\n\nGrammatical coding varies with audience. Specialist works may use compact category systems, while learner’s dictionaries often give patterns and examples in less technical form. Noun entries can distinguish count and mass uses: a chicken and some chicken differ in interpretation as well as grammatical behavior. Adjective entries may record restrictions on attributive or predicative position. Such information links lexical description to grammar without requiring the dictionary to reproduce an entire grammatical reference work.\n\nPronunciation can be represented through phonetic transcription, respelling systems, audio, or combinations of these. The International Phonetic Alphabet provides a widely used framework, though individual dictionaries make different decisions about phonemic detail and accent coverage. A transcription must specify which pronunciation variety it represents. Stress, vowel reduction, and variants can be especially important where spelling gives an uncertain guide to sound.\n\nAudio adds evidence of realization but does not remove editorial choices. A recording represents a particular speaker, accent, speaking style, and production context. Digital dictionaries may provide several regional models or distinguish isolated citation forms from connected-speech variants. For sign languages, video is often central because lexical forms involve handshape, location, movement, orientation, and nonmanual features. Access may be organized through these properties rather than through an alphabetical sequence of spoken-language glosses.\n\nPhraseology concerns conventional combinations larger than a single word. Dictionaries must distinguish freely assembled phrases from collocations, idioms, and relatively fixed constructions. Strong tea illustrates a preferred combination that is not fully predictable from meaning alone; kick the bucket requires an idiomatic interpretation in appropriate contexts. Whether such expressions receive independent entries or appear under a component word affects how easily users can locate them.\n\nUsage labels locate expressions socially, geographically, historically, or stylistically. Labels such as informal, regional, dated, and offensive convey different kinds of information and should not be treated as interchangeable. Register concerns associations with situations and activities, while regional labels identify distribution rather than inferiority. Usage notes can distinguish disputed conventions from well-established grammatical facts. Clear policies are particularly important where an expression is grammatical but inappropriate in some social circumstances."
    },
    {
      "heading": "Bilingual and multilingual lexicography",
      "text": "Bilingual lexicography links lexical resources across languages, but it cannot generally replace each source-language word with one target-language word. Translation equivalence depends on sense, context, grammatical construction, and communicative purpose. English know, for example, corresponds to different French verbs in constructions involving knowledge of a fact and familiarity with a person. An entry must therefore discriminate among uses before proposing equivalents.\n\nThe direction of consultation matters. A dictionary intended to help English speakers read French need not contain the same information as one intended to help French speakers write English. Productive use requires guidance about target-language grammar, collocation, and stylistic restrictions. Receptive use can often begin with a compact equivalent or explanatory gloss. A physically bidirectional volume may consequently contain two sections with different information needs rather than two mechanically reversed lists.\n\nLexical mismatches arise because languages divide conceptual domains differently. Kinship terms may encode distinctions that English leaves to phrases; motion verbs may distribute manner and direction differently; administrative titles may refer to institutions with no direct counterpart. A suitable treatment can combine an approximate equivalent with an explanation and examples. A borrowing or descriptive phrase may be more accurate than a deceptively familiar term.\n\nEvidence from a parallel corpus helps reveal how translators handle particular expressions. However, an observed translation is not automatically an interchangeable equivalent. Translators may omit, paraphrase, restructure, or adapt material, and the appropriate solution may depend on a whole passage. Comparable corpora—independent texts from similar domains in different languages—can supplement parallel evidence by showing conventional formulations in each language.\n\nMultilingual projects face the problem of maintaining consistent distinctions across more than two languages. A concept-centered resource can organize language-specific designations around shared subject categories, particularly in technical domains. General vocabulary is less easily aligned because lexical distinctions do not always converge on a neutral set of concepts. Forcing every language into the sense divisions of one pivot language can conceal significant differences.\n\nScript and access design introduce further requirements. Transliteration preserves systematic relations between writing systems, while pronunciation-oriented transcription serves another purpose. Dictionaries may need both native-script forms and a recognized romanization system, together with cross-references for spelling variation. Resources for languages with limited written documentation may combine dictionary compilation with audio recording and community review. In these settings, translation equivalents, grammatical analysis, and locally meaningful explanations often have to be developed together rather than transferred from an established external dictionary."
    },
    {
      "heading": "Digital systems and computational lexicography",
      "text": "Digital lexicography separates the underlying lexical database from the interface through which a user consults it. A single structured entry can generate multiple presentations, including a concise mobile result, a full historical account, and machine-readable data. This separation enables correction and reuse, but only if the database distinguishes senses, forms, examples, labels, and citations consistently. Digitizing a printed page does not by itself produce a fully structured lexical resource.\n\nSearch changes access without eliminating the need for editorial organization. An online system can accept inflected forms, common misspellings, pronunciation approximations, or phrases containing unknown components. Results may be ranked by frequency, query context, or user-selected language. These conveniences require explicit links among forms and entries. They also create risks: an apparently relevant result can conceal that the query matched a different sense or a merely similar spelling.\n\nStandards support exchange among systems. The Text Encoding Initiative provides guidelines used to encode dictionary content, and the Lexical Markup Framework, standardized as ISO 24613, provides a framework for modeling lexical resources. Stable identifiers help preserve links when definitions or headwords change. An application programming interface can supply selected content to writing tools, educational applications, and other services, subject to licensing and technical restrictions.\n\nComputational lexical resources are related to, but not identical with, dictionaries for human consultation. WordNet, developed at Princeton beginning in the 1980s, organizes lexical items into sets of cognitive synonyms and records semantic relations among them. Such networks support computational tasks, but their sense inventories and definitions reflect particular design goals. Converting a human-oriented dictionary into a computational resource requires resolving abbreviations, implicit relationships, and presentation conventions that readers ordinarily infer.\n\nMachine learning can assist with candidate-word detection, example selection, semantic clustering, and identification of changing usage. A large language model can also generate draft explanations or propose distinctions for review. Fluent output is not evidence that a definition is accurate or that a quotation genuinely occurred. Lexicographic use therefore depends on connecting proposed descriptions to inspectable sources and checking them against the project’s policies.\n\nDigital publication permits frequent updates, but continuous availability should not be confused with continuous completeness. Historical entries may be revised in batches, and specialist sections may progress at different rates. Version dates, revision histories, and citation information help users interpret a changing resource. For long-term preservation, editorial teams must also maintain export formats, backups, and migration plans; a dictionary can remain intellectually valuable after the software originally used to produce it has become obsolete."
    },
    {
      "heading": "Authority, representation, and editorial responsibility",
      "text": "Dictionaries acquire authority through editorial consistency, breadth of evidence, institutional reputation, and habitual use. That authority is often interpreted more strongly than their methods justify. Inclusion normally means that an expression meets a resource’s coverage criteria, not that editors approve of it. Conversely, exclusion can result from limited space, insufficient evidence, delayed revision, or a restricted scope. These distinctions are central to understanding the relationship between dictionaries and a standard language.\n\nDescription and prescription are not completely separable practices. A descriptive dictionary reports attested patterns, yet it must choose a spelling for its headword, select pronunciation models, and decide which labels to attach. A prescriptive reference work still needs evidence to identify actual variation and recurring uncertainty. The significant questions are what kind of authority is claimed, which linguistic community is represented, and whether editorial judgments are distinguished from empirical observations.\n\nRepresentation depends heavily on the source record. Written corpora often preserve institutional and published language better than informal conversation, minority varieties, or historically marginalized speakers. A dictionary based on those sources can reproduce their imbalance even when individual entries are carefully researched. Regional corpora, oral documentation, and collaboration with speech communities can expand the evidence base. These additions require attention to consent, attribution, and the circumstances under which recordings or texts were collected.\n\nOffensive vocabulary presents a distinct descriptive task. Historical and contemporary dictionaries may need to document such expressions so that users can understand texts or recognize harmful language. Appropriate treatment distinguishes the expression’s meaning, its targets, its social force, and its historical distribution. Definitions should not repeat the prejudice as an unqualified statement of fact. Examples and quotations must be selected for explanatory value rather than simply reproduced because they occur in a source.\n\nCopyright and access conditions shape both production and circulation. Definitions, selected examples, database organization, and software may involve different intellectual property interests, depending on the jurisdiction and the material concerned. Openly licensed resources facilitate reuse and correction, while subscription and licensing models can finance sustained editorial work. Neither model alone guarantees accuracy, comprehensive coverage, or long-term preservation.\n\nQuality assessment consequently extends beyond counting headwords. It includes the reliability of evidence, consistency of sense distinctions, accuracy of grammatical information, clarity of labels, accessibility of the interface, and success in representative consultation tasks. Corrections submitted by users can identify omissions or mistakes, but reports require editorial assessment. In collaborative resources, transparent revision histories and explicit contribution policies make it possible to examine how individual lexical claims entered the published record."
    }
  ],
  "links": [
    "linguistics",
    "translation",
    "lexeme",
    "cuneiform",
    "writing-system",
    "latin",
    "printing",
    "academie-francaise",
    "collocation",
    "lemmatization",
    "inflection",
    "chinese-radicals",
    "pinyin",
    "corpus-linguistics",
    "text-corpus",
    "concordance",
    "part-of-speech-tagging",
    "syntactic-parsing",
    "polysemy",
    "metaphor",
    "morphology",
    "syntax",
    "international-phonetic-alphabet",
    "sign-language",
    "register-linguistics",
    "romanization",
    "application-programming-interface",
    "machine-learning",
    "large-language-model",
    "standard-language",
    "intellectual-property"
  ],
  "sources": [],
  "license": "https://creativecommons.org/licenses/by/4.0/"
}