This document specifies a set of Label Generation Rules (LGR) for the Latin script for the second level domain or domains identified above. This LGR contains the full set of cross-script variants required for use with the following scripts: Armenian, Cyrillic, and Greek, plus Hebrew. The starting point for the development of this LGR can be found in the related Root Zone LGR [RZ-LGR-Latn]. The format of this file follows [RFC 7940]. This LGR is adapted from the “Reference LGR for the Second Level for the Latin Script” [Ref-LGR-und-Latn], for details, see Change History below.
For details and additional background on the Latin script, see “Proposal for a Latin Script Root Zone Label Generation Rule-Set (LGR)” [Proposal-Latin].
The repertoire contains the 197 letters needed to write hundreds of languages in the Latin script. An additional 7 combining diacritical marks are available as part of 21 explicitly defined combining sequences. The repertoire is a subset of [Unicode 11.0.0]. For details, see Section 5, “Repertoire” in [Proposal-Latin]. (The proposal cited has been adopted for the Latin script portion of the Root Zone LGR.)
Compared to that source, an additional language is supported by adding the code point for the Middle Dot used in the Catalan Ela Geminada: U+006C U+00B7 U+006C. Context rules limit U+00B7 MIDDLE DOT to being bracketed by the letter “l”. (See also [280])
For the second level, the repertoire has been augmented with the ASCII digits, U+0030 to U+0039, plus U+002D HYPHEN-MINUS, for a total of 239 repertoire elements.
Any code points outside the Latin Script repertoire that are targets for out-of-repertoire variants would be included here only if the variant is listed in this file. In this case they are identified as a reflexive (identity) variant of type “out-of-repertoire-var”. Whether or not they are listed, they do not form part of the repertoire.
Repertoire Listing: Each code point or range is tagged with the script or scripts with which the code point is used and one or more other character categories. For each repertoire element, one or more references document sufficient justification for inclusion in the repertoire; see the “References” below. For code points that are part of the repertoire, comments identify the languages using the code point along with their [EGIDS] level.
The Latin script is a major writing system of the world, and the most widely used in terms of the number of languages and speakers, with circa 70% of the world’s readers and writers making use of this script. From a list of 1,189 languages using the Latin script [Omniglot] the 212 languages that were taken into consideration contain all 182 languages with [EGIDS] level 1–4 together with many languages with EGIDS level 5, each spoken by more than 1 million estimated speakers. Altogether over 100 languages are cited here to justify specific additions to the repertoire, but many other languages may also be written using some subset of the repertoire of this LGR. In a few cases, code points were excluded in [MSR-5] due to security concerns; for the affected languages, only a subrepertoire could be supported. More details in Section 3, “Background on Script and Principal Languages Using It” of [Proposal-Latin].
Unlike other LGRs for this script, which only incude the variant definitions needed when used as standalone in their zone, this LGR contains the full set of variants needed for zones shared by more than one script. Where appropriate, cross-script variants have been defined to mutually exclude labels from different scripts that could otherwise be substituted by the users. To meet the requirement for variant transitivity, additional in-script variants may be defined as needed. Once a label has been delegated for one script, any of its variant labels consisting entirely of cross-script variants would be blocked. Any label containing at least one code point that is not a cross-script variant would be considered distinct. Because of that, even large numbers of defined cross-script variants generally do not lead to a high percentage of labels experiencing a collision across scripts.
The definition of variants included in this LGR contributes to the definition of variants in the Common LGR. Even though this LGR contains the full set of variants affecting this script, use of the Common LGR is required for actual mitigation of issues caused by using multiple languages or scripts in the same zone. For details, see Section 3, "Use of Multiple Reference LGRs in the Same Zone" in [Level-2-Overview].
In-script and cross-scripts variants defined in this LGR largely follow the methodology defined in Section 6, “Variants”, in [Proposal-Latin]. In a separate appendix that proposal identifies additional candidates for visually confusable code points (see [Proposal-Latin-Appendices]). They provide data on the basis of which additional variants, or fallback variants (see following) might be defined. However, doing so would require additional review and analysis that has not been carried out for this version of the LGR.
The LGR defines certain allocatable fallback variants as described in Section 4.5.5 “Allocatable Fallback Variants” in [Level-2-Overview]. A fallback variant is a variant label that uses substitute code points for code points or sequences not available (or not allowed) in some contexts, that would otherwise be required for a linguistically accurate rendering of some label.
When “fallback” variants are defined, two labels may be allocated: a single label with the spelling preferred by the applicant, plus a single fallback variant for that label. The fallback exclusively uses the fallback characters for any characters for which fallbacks are defined, while the “preferred” label may use any otherwise valid mix of code points. If the fallback variant is the one applied for, no other variant label is allocatable.
An allocatable fallback variant exists for the following pairs where the second element of each pair is the fallback:
U+00B7 MIDDLE DOT and U+002D HYPHEN-MINUS — the use of the hyphen as fallback for the middle dot in the Catalan Ela Geminada follows registry practice, see [281]. The variant is limited to an Ela Geminada context.
Some second level LGRs provide ASCII fallback variants for some or all accented Latin characters. Likewise the U+0153 Small OE Ligature and U+00E6 Small AE ligature have ASCII fallbacks consisting of the non-ligated “oe” and “ae” sequences. None of these fallbacks have been added to the current version of the LGR.
Overlapped Variant Sequence: Both “ss” and “s” coexist in the repertoire and “s” has variant relationships on its own. These variants thus overlap: making the variant set well-behaved for index variant calculation requires that the sequence “ss” also be given variants to all permutations of variants for the letter s followed by itself, as well as all transitive variants due to other variants for U+00DF, including out-of-repertoire sequences.
In each of the fallback variant pairs defined above, the mapping type from the first element to the second is of type “fallback”, while the variant type for the other direction is “blocked”. In addition, the first element of each pair uses the reflexive mapping “r-original”. (By convention, the prefix “r-” marks a type used in a reflexive variant mapping, that is, it represents an instance of the original code point at that location in a variant label, see Section 5.3.4 in [RFC 7940].)
Where homoglyphs or near homoglyphs exist between the Latin, Greek, Cyrillic and Armenian scripts, out-of-repertoire variants are defined with a comment “Cross-script homoglyph” or “Cross-script near homoglyph” respectively. As result of integration, including the effects of cross-script transitivity, the number of cross-script variants listed for these scripts may be larger than defined in Section 6, “Variants” in [Proposal-Latin].
This LGR lists in full the variant mappings among the closely related scripts Armenian, Cyrillic, and Greek, plus Hebrew. Some other scripts also have variants of ASCII characters, but these are listed elsewhere. In addition, the Latin LGR inherits a number in-script variants as result of integration, primarily due to variants with the Greek script. These variants are marked in comments as “Required for integration”. Some of these might have been re-analyzed as fallback variants, but this has not been carried out in this version of the LGR.
Implicit Variant Sequences: The full set of inherited cross script variants would include certain otherwise redundant sequences. These are defined as having a partition into other repertoire elements, where each element has a cross-script variant of its own, making a concatenation of these variants an implicit cross-script variant for the sequence. Where such sequences do not contribute to the set of available valid labels, where they produce the same index variant as all of their valid partitions, and where they do not participate in allocatable variants, they are not listed anywhere. See Section 4.5.8 “Implicit Cross-script Variants” of [Level-2-Overview] for further details.
Note that U+0073 U+0073 has cross script variants that are not listed, but is itself target of a “fallback” variant mapping potentially leading to an allocatable label.
In addition, a number of letter forms like “i”, “o” and “c” are rather generic in appearance and thus give rise to cross-script variant relations with otherwise unrelated scripts.
Variant Disposition: Except for limited exceptions for the fallback variants defined above, variants defined here result in a variant label disposition of “blocked”.
For more details, see Section 11.0.2 “Cross-Script Variants” in [Proposal-Latin].
The specification of variants in this LGR follows the guidelines in [RFC 8228].
This LGR does not define named character classes.
By default, the LGR includes the rules and actions to implement the following restrictions mandated by the IDNA protocol. They are marked with ⍟.
The following context rule applies to U+00B7 MIDDLE DOT and its variants. It ensures that the middle dot is part of an Ela Geminada sequence and variants between it and HYPHEN-MINUS are only defined in that context.
The following WLE rule invalidates labels in which two Ela Geminada sequences overlap.
This LGR includes the default actions for LGRs as well as the action needed to invalidate labels with misplaced combining marks. They are marked with ⍟. For a description see [RFC 7940].
Because this LGR defines allocatable fallback variants the following default actions are applicable.
These actions resolve as “allocatable” any label where all variants are of type “fallback”, and as “valid” any label where all variants are of type “r-original”. Labels with a mix of variant types are resolved as “blocked”.
To account for original code points in a permuted variant, reflexive variant mappings with an “r-” prefix are used. (See [RFC 7940]). In particular, the mapping type “r-original” is given to any code point that has a fallback mapping, but that appears in its non-fallback form in the original label, and thus “maps to itself”.
Default actions that are triggered by the LGR-specific variant types described above limit the “allocatable” variant labels to those containing only “ss”, dotted “i” or hyphen variants, while disallowing mixed use of “ss” and “ß”, Dotless i and “i”, or middle-dot and hyphen respectively, except as in the original applied-for label.
Note that the mapping types for variants are not symmetric: they depend on which code point is considered the source or the target in a given mapping. As specified in [RFC 7940], when mapping types are evaluated code points in a label that are unchanged use the type of their “reflexive” mapping. Per [RFC 7940] the actions are always applied one after the other, and the evaluation stops at the first action that assigns a disposition to a given label.
An action has been defined to invalidate labels containing the sequence U+00B7 U+006C U+00B7 which could otherwise result in two Ela Geminada sequences overlapping.
The LGR in this document has been adapted from the corresponding Reference LGR for the Second Level. The Second Level Reference LGR for the Latin Script was developed by Michel Suignard and Asmus Freytag, based on the Root Zone LGR for the Latin script and information contained or referenced therein; see [RZ-LGR-Latn]. Suitable extensions for the second level have been applied according to the [Guidelines] and with community input. The original proposal for a Root Zone LGR for the Latin script, that this LGR is based on, was developed by the Latin Generation Panel. For more information on methodology and contributors to the underlying Root Zone LGR, see Sections 4 and 8 in [Proposal-Latin], as well as [RZ-LGR-Overview].
This version modifies the Second Level Reference LGR for the Latin Script [Ref-LGR-und-Latn] by injecting the complete (maximal) set of cross-repertoire variants, including any imposed by transitivity. Injected variants are marked by special comments. Any character classes, context and WLE rules required for testing are also imported. In all other respects, description or annotations have not been updated to reflect the injected variants. Any LGR elements that might pose a possible conflict when the LGR is used with the Common LGR for cross-repertoire variant resolution are clearly marked. All modifications are carried out mechanically; there may be an occasional context rule or similar that is injected because its definition relates to some repertoire element even though it can never be triggered by any label valid under the modified LGR.
The following general references are cited in this document:
For references consulted particularly in designing the repertoire for the Latin Script for the second level please see details in the Table of References below.
References [0] and up refer to the Unicode Standard versions in which corresponding code points were initially encoded. References [101] and up correspond to a source given in [Proposal-Latin] for justifying the inclusion of the corresponding code points. In the listing of whole label evaluation and context rules, reference [320] indicates the source for common rules.
]]>