Nickmancol commited on
Commit
bddf16c
·
1 Parent(s): 272c554

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +74 -1
README.md CHANGED
@@ -7,4 +7,77 @@ tags:
7
  pretty_name: Etymology DB
8
  size_categories:
9
  - 1M<n<10M
10
- ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7
  pretty_name: Etymology DB
8
  size_categories:
9
  - 1M<n<10M
10
+ ---
11
+
12
+ # etymology-db
13
+
14
+ This is just a stub from the original [repo](https://github.com/droher/etymology-db)
15
+ All credits go to the original author [David Roher](https://github.com/droher).
16
+
17
+ A structured, comprehensive, and multilingual etymology dataset created by parsing Wiktionary's etymology sections. Key features:
18
+ * 4.2+ million etymological relationships between 2.0+ million terms in 3300+ languages/dialects
19
+ * 31 different types of etymological relations, distinguishing between inheritance, borrowing, etc.
20
+ * Hierarchical data that preserves relationship structures, such as the evolution of a term across languages
21
+
22
+ Caveat for people interested in using this for research: all information is pulled directly from Wiktionary via
23
+ semi-structured text parsing, and I've made no effort yet to validate any particular result. That said, I would love
24
+ for this to be useful, so please raise an issue if you have questions.
25
+
26
+ Here is a description of the table schema:
27
+
28
+
29
+ | Column Name | Description |
30
+ |-----------------|----------------------------------------------------------------------------------------------------------------------------------------|
31
+ | term_id | A hash of the term and its language. |
32
+ | lang | The language/dialect of the term. |
33
+ | term | The term itself. Usually a word, but can also be a prefix or a multi-word expression, hence "term" instead of word. |
34
+ | reltype | The kind of etymological relation being specified (see below for details on each possible value). |
35
+ | related_term_id | A hash of the related term and its language (useful for assembling relationships across multiple terms). |
36
+ | related_lang | The language/dialect of the related term. NULL for parent root nodes. |
37
+ | related_term | The term that is etymologically related to the original entry. NULL for parent root nodes. |
38
+ | position | Zero-indexed position of the term when the relation is made up of multiple terms (e.g. a compound). |
39
+ | group_tag | Randomly generated ID. populated only for the root nodes of nested relationships. |
40
+ | 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. |
41
+ | parent_position | Zero-indexed position of the relation inside of its nested structure. NULL if not nested. |
42
+
43
+ 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.
44
+
45
+ | Relation Type | Description |
46
+ |-----------------------------|------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
47
+ | inherited_from | Indicates that `term` has an unbroken chain of inheritance from `related_term`. |
48
+ | borrowed_from | Indicates that `term` is a loanword borrowed during the time the borrowing language was spoken. |
49
+ | derived_from | A catch-all for a derivation relationship that is not specifically inherited/borrowed. |
50
+ | learned_borrowing_from | Borrowed from the original language via atypical ("inorganic") means of language contact. |
51
+ | semi_learned_borrowing_from | Borrowed words that have been partly reshaped by later sound change or analogy with inherited terms. /
52
+ | orthographic_borrowing_from | Borrows the spelling of `related_term` but not the pronunciation. |
53
+ | unadapted_borrowing_from | Borrowed words that have not been conformed to the morpho-syntactic, phonological and/or phonotactical rules of the target language. |
54
+ | root | Constructed root(s) of `term` in a theoretical ancestor language, e.g. Proto-Indo-European. |
55
+ | has_prefix | Indicates that `term` is partially based on the prefix `related_term`. |
56
+ | has_prefix_with_root | `related_term` is the term attached to a prefix (not necessarily a suffix, e.g. "normal" in "abnormal") |
57
+ | has_suffix | Same as above, but for suffixes. |
58
+ | has_suffix_with_root | Same as above, but for suffixes. |
59
+ | 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. |
60
+ | has_affix | Affix is the general form of prefix/suffix/confix and indicates some kind of compound structure without further detail. |
61
+ | compound_of | Indicates that `related_term` is the `position`-indexed term of a compound that makes up `term`. Used interchangeably with affix above. |
62
+ | back-formation_from | Indicates that `term` was formed from `related_term` by removing a prefix/suffix. |
63
+ | doublet_with | Indicates that `term` and `related_term` have the same etymological origin, especially when the relationship is unintuitive. |
64
+ | is_onomatopoeic | Indicates that `term` is an onomatopoeia (a word form from a sound associated with its meaning). |
65
+ | calque_of | Indicates that `term` is borrowed from `related_term` via a direct word-for-word or root-for-root translation. |
66
+ | semantic_loan_of | Special case of calques in which the word already existed but a new meaning was added. |
67
+ | named_after | Indicates that `term` is based on the name of the person `related_term` (an eponym). |
68
+ | phono-semantic_matching_of | Indicates that `term` and `related_term` have very similar sounds and meanings in both languages. |
69
+ | etymologically_related_to | A catch-all indicating `term` and `related_term` are etymologically related without any further context provided. |
70
+ | blend_of | Indicates that `term` is made up of a blend of `related_term` and the related terms in other `position`s. This differs from a compound in that the beginning of one word is combined with the ending of another. |
71
+ | clipping_of | Indicates that `term` is a _spoken_ shortened version of `related_term` without any semantic difference. |
72
+ | abbreviation_of | Indicates that `term` is a _written_ shortened version of `related_term` without any semantic difference. |
73
+ | initialism_of | Indicates that `term` is based on the initials of `related_term`. |
74
+ | cognate_of | Indicates that `term` and `related_term` sound/mean similar things, but no direct ancestral relationship exists. |
75
+ | group_affix_root | A node that groups together rows that, when combined, form an affix. |
76
+ | 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. |
77
+ | group_derived_root | A node that groups together rows that, when combined, form an unbroken chain of inheritance (in reverse chronological order). |
78
+
79
+ The `wiktionary_codes.csv` file is manually combined from these two pages:
80
+ https://en.wiktionary.org/wiki/Wiktionary:List_of_languages
81
+ https://en.wiktionary.org/wiki/Module:etymology_languages/data
82
+
83
+ All data is licensed under the Creative Commons ShareAlike 3.0 License. All code is licensed under the Apache 2.0 license.