Downloads: (Last generated 2021-11-14)
Gzipped CSV
Parquet
A structured, comprehensive, and multilingual etymology dataset created by parsing Wiktionary's etymology sections. Key features:
- 3.8+ million etymological relationships between 1.8+ million terms in 2900+ languages/dialects
- 31 different types of etymological relations, distinguishing between inheritance, borrowing, etc.
- Hierarchical data that preserves relationship structures, such as the evolution of a term across languages
Caveat for people interested in using this for research: all information is pulled directly from Wiktionary via semi-structured text parsing, and I've made no effort yet to validate any particular result. That said, I would love for this to be useful, so please raise an issue if you have questions.
Here is a description of the table schema:
| Column Name | Description |
|---|---|
| term_id | A hash of the term and its language. |
| lang | The language/dialect of the term. |
| term | The term itself. Usually a word, but can also be a prefix or a multi-word expression, hence "term" instead of word. |
| reltype | The kind of etymological relation being specified (see below for details on each possible value). |
| related_term_id | A hash of the related term and its language (useful for assembling relationships across multiple terms). |
| related_lang | The language/dialect of the related term. NULL for parent root nodes. |
| related_term | The term that is etymologically related to the original entry. NULL for parent root nodes. |
| position | Zero-indexed position of the term when the relation is made up of multiple terms (e.g. a compound). |
| group_tag | Randomly generated ID. populated only for the root nodes of nested relationships. |
| parent_tag | If this relation is inside of a nested structure, this will be populated with the group_tag of its immediate parent. NULL otherwise. |
| parent_position | Zero-indexed position of the relation inside of its nested structure. NULL if not nested. |
And here is a description of each relation type. Note that these are all derived directly from Wiktionary's etymology templates -- all rows are classified according to the name of the template from which the info was extracted, and no further inferences are made. The only exception to this is the group relations, which are based on formulaic and reoccuring patterns in natural language sections.
| Relation Type | Description |
|---|---|
| inherited_from | Indicates that term has an unbroken chain of inheritance from related_term. |
| borrowed_from | Indicates that term is a loanword borrowed during the time the borrowing language was spoken. |
| derived_from | A catch-all for a derivation relationship that is not specifically inherited/borrowed. |
| learned_borrowing_from | Borrowed from the original language via atypical ("inorganic") means of language contact. |
| semi_learned_borrowing_from | Borrowed words that have been partly reshaped by later sound change or analogy with inherited terms. / |
| orthographic_borrowing_from | Borrows the spelling of related_term but not the pronunciation. |
| unadapted_borrowing_from | Borrowed words that have not been conformed to the morpho-syntactic, phonological and/or phonotactical rules of the target language. |
| root | Constructed root(s) of term in a theoretical ancestor language, e.g. Proto-Indo-European. |
| has_prefix | Indicates that term is partially based on the prefix related_term. |
| has_prefix_with_root | related_term is the term attached to a prefix (not necessarily a suffix, e.g. "normal" in "abnormal") |
| has_suffix | Same as above, but for suffixes. |
| has_suffix_with_root | Same as above, but for suffixes. |
| has_confix | A confix is a term whose first element is a prefix and whose last is a suffix. Position 0 of a confix is the prefix, and the last position is the suffix. |
| has_affix | Affix is the general form of prefix/suffix/confix and indicates some kind of compound structure without further detail. |
| compound_of | Indicates that related_term is the position-indexed term of a compound that makes up term. Used interchangeably with affix above. |
| back-formation_from | Indicates that term was formed from related_term by removing a prefix/suffix. |
| doublet_with | Indicates that term and related_term have the same etymological origin, especially when the relationship is unintuitive. |
| is_onomatopoeic | Indicates that term is an onomatopoeia (a word form from a sound associated with its meaning). |
| calque_of | Indicates that term is borrowed from related_term via a direct word-for-word or root-for-root translation. |
| semantic_loan_of | Special case of calques in which the word already existed but a new meaning was added. |
| named_after | Indicates that term is based on the name of the person related_term (an eponym). |
| phono-semantic_matching_of | Indicates that term and related_term have very similar sounds and meanings in both languages. |
| etymologically_related_to | A catch-all indicating term and related_term are etymologically related without any further context provided. |
| blend_of | Indicates that term is made up of a blend of related_term and the related terms in other positions. This differs from a compound in that the beginning of one word is combined with the ending of another. |
| clipping_of | Indicates that term is a spoken shortened version of related_term without any semantic difference. |
| abbreviation_of | Indicates that term is a written shortened version of related_term without any semantic difference. |
| initialism_of | Indicates that term is based on the initials of related_term. |
| cognate_of | Indicates that term and related_term sound/mean similar things, but no direct ancestral relationship exists. |
| group_affix_root | A node that groups together rows that, when combined, form an affix. |
| group_related_root | A node that groups together rows in which related_terms are not just related to the term, but to each other as well. |
| group_derived_root | A node that groups together rows that, when combined, form an unbroken chain of inheritance (in reverse chronological order). |
The wiktionary_codes.csv file is manually combined from these two pages:
https://en.wiktionary.org/wiki/Wiktionary:List_of_languages
https://en.wiktionary.org/wiki/Module:etymology_languages/data
All data is licensed under the Creative Commons ShareAlike 3.0 License. All code is licensed under the Apache 2.0 license.