Elamur: **Mova** · [Parse](https://mova.elamur.ai/free/en/parse/) · [Dictionary](https://md.dictionary.elamur.ai/free/en/) · [Translate](https://md.translate.elamur.ai/free/en/) · [Research](https://md.research.elamur.ai/free/en/) · [Gali](https://md.gali.elamur.ai/free/en/) · [MCP](https://mcp.elamur.ai/) · [Locations](https://md.locations.elamur.ai/free/en/) · [Weather](https://md.weather.elamur.ai/free/en/)

[Mova](https://md.mova.elamur.ai/free/en/) · [this page in HTML](https://mova.elamur.ai/free/en/docs/languages/ku/ckb/01-script-and-unicode/)

[Mova](https://md.mova.elamur.ai/free/en/) / [docs](https://md.mova.elamur.ai/free/en/docs/) / [languages](https://md.mova.elamur.ai/free/en/docs/languages/) / [ku](https://md.mova.elamur.ai/free/en/docs/languages/ku/) / [ckb](https://md.mova.elamur.ai/free/en/docs/languages/ku/ckb/) / 1. The Sorani script

# 1. The Sorani script

**On this page**

1. [The Arabic-derived alphabet and joining behavior](#the-arabic-derived-alphabet-and-joining-behavior)
2. [Consonant letters and letters with multiple readings](#consonant-letters-and-letters-with-multiple-readings)
3. [Vowel letters and the ordinarily unwritten short vowel](#vowel-letters-and-the-ordinarily-unwritten-short-vowel)
4. [Initial vowel carrier and optional marks](#initial-vowel-carrier-and-optional-marks)
5. [Unicode characters and visually similar letter forms](#unicode-characters-and-visually-similar-letter-forms)
6. [Right-to-left text and mixed-direction passages](#right-to-left-text-and-mixed-direction-passages)
7. [Romanization and its relation to pronunciation](#romanization-and-its-relation-to-pronunciation)
8. [How Mova handles the Sorani script](#how-mova-handles-the-sorani-script)
9. [Sources for this chapter](#sources-for-this-chapter)

Central Kurdish is ordinarily written in an Arabic-derived alphabet, from right to left. The alphabet gives separate written forms to several Kurdish consonants and makes most vowels explicit. Short *i* is ordinarily unwritten, and two letters serve both as consonants and as vowels. Reading therefore involves identifying the letters, deciding which sound values they have in the word, and supplying any unwritten short vowels. This chapter uses a single romanization throughout and identifies the points where a romanized reading contains more information than the spelling. (Ahmadi 2020, pp. 72–84)

## The Arabic-derived alphabet and joining behavior

Sorani shares many letter shapes with Arabic and Persian. Its inventory includes forms such as `پ`, `چ` and `گ`, as well as `ڕ`, `ڵ`, `ۆ`, `ێ` and `ە`, which carry distinctions used in Kurdish. Letters imported with the script have Kurdish sound values and spelling conventions. Soane's account of Kurdish writing records the need to assign Kurdish values to Arabic-derived letters. His description covers more than one variety and uses a historical romanization. The tables below give the readings used for present-day Sorani in this chapter. (Soane 1913, pp. 20, 23; Ahmadi 2020, pp. 72–84)

### How letters join

The letter at the start of a Sorani word is on the right. In the table, *preceding* and *following* refer to reading order. A preceding Sorani letter is normally to the right of the current letter. Two adjacent letters join only when the earlier letter can join forward and the later letter can join back.

| Joining class | Letters in the chapter's alphabet | Join within the word |
| --- | --- | --- |
| Dual-joining | `ئ ب پ ت ج چ ح خ س ش ع غ ف ڤ ق ک گ ل ڵ م ن ه ھ ی ێ` | Can join a preceding or a following letter when that neighbor permits the connection |
| Right-joining | `ا د ر ڕ ز ژ ە و ۆ` | Can join a preceding letter when it permits the connection. A connection to the following letter is blocked |

A dual-joining letter can have an isolated, initial, medial or final shape. The shape describes the connections actually present. An initial shape connects forward, a final shape connects back, and a medial shape connects in both directions. A letter with an initial shape can occur inside a word after a right-joining letter. A right-joining letter has isolated and final shapes. The distinct shapes used in ordinary typesetting are contextual renderings of the same character.

For example, `چەقۆ` contains two connected groups of letters within one word. `چ` joins forward to `ە`, where the first group ends. `ق` then starts a new group and joins forward to `ۆ`. In `دوو`, `د` and both instances of `و` are right-joining, so all three letters appear without connections. Spaces are written at word boundaries according to the spelling of the text. A gap produced by a right-joining letter is part of the shape of the word. (Ahmadi 2020, pp. 72–84)

| Form | Romanization | Gloss | Translation | Corpus sentence ID |
| --- | --- | --- | --- | --- |
| چەقۆ | çeqo | knife | “knife” | `wiki:ckb:چەقۆ` |
| دوو | dû | two | “two” | `wiki:ckb:چەقۆ` |
| ماڵپەڕی فەرمیی | małpeř-î fermî-y | website-EZ official-EZ | “the official website of” | `wiki:ckb:ڕێکخراوەی ھاوکاریی شانگھای` |

In the last example, `ڕ` ends a connected group within `ماڵپەڕی` even though the ezafe letter follows in the same word. `EZ` in the gloss marks ezafe. Hyphens in the romanization and gloss identify grammatical boundaries for the reader.

## Consonant letters and letters with multiple readings

The consonant table lists the ordinary written consonants, including the consonantal readings of `و` and `ی`. The IPA column gives broad values. The pronunciation of some sounds, particularly sounds found in loans, varies by speaker and variety. `ئ` is treated with the vowel carrier below. (Ahmadi 2020, pp. 72–84)

| Letter | Romanization | Broad consonant value |
| --- | --- | --- |
| `ب` | b | /b/ |
| `پ` | p | /p/ |
| `ت` | t | /t/ |
| `ج` | c | /dʒ/ |
| `چ` | ç | /tʃ/ |
| `ح` | ḥ | /ħ/ |
| `خ` | x | /x/ |
| `د` | d | /d/ |
| `ر` | r | /ɾ/, a tap |
| `ڕ` | ř | /r/, a trill |
| `ز` | z | /z/ |
| `ژ` | j | /ʒ/ |
| `س` | s | /s/ |
| `ش` | ş | /ʃ/ |
| `ع` | ʿ | /ʕ/ in pronunciations that retain the pharyngeal consonant |
| `غ` | gh | /ɣ/ |
| `ف` | f | /f/ |
| `ڤ` | v | /v/ |
| `ق` | q | /q/ |
| `ک` | k | /k/ |
| `گ` | g | /ɡ/ |
| `ل` | l | /l/ |
| `ڵ` | ł | /ɫ/, a velarized lateral |
| `م` | m | /m/ |
| `ن` | n | /n/ |
| `ه`, `ھ` | h | /h/ in their consonant use |
| `و` | w | /w/ in its consonant use |
| `ی` | y | /j/ in its consonant use |

The letters `ر` and `ڕ` distinguish two rhotics. The letters `ل` and `ڵ` distinguish two laterals. Preserve these distinctions in transcription even where the contrasts are unfamiliar to an English reader. In `شەڕ`, the rhotic is `ڕ`. In `ماڵپەڕی`, the lateral is `ڵ` and the rhotic is `ڕ`. In `ڤیزە`, the initial letter is `ڤ` and the medial `ی` writes a vowel. Soane's description distinguishes lateral qualities and records consonant sounds at different places of articulation. His printed symbols and dialect comments belong to his own system. (Soane 1913, p. 23)

`و` can stand for consonantal *w* or vocalic *u*. The usual spelling of *û* uses two consecutive instances of `و`. The consonant in `وێژە` is apparent from the following vowel `ێ`. The word `دوو` illustrates the vowel digraph `وو`. Decide the reading in the word as a whole. In particular, the presence of a neighboring consonant letter does not tell the reader whether there is an unwritten vowel between the two.

`ی` can stand for consonantal *y* or vocalic *î*. Before a written vowel it can supply a syllable onset, as in `کایە`. In `تیر` it supplies the vowel. At a grammatical boundary it can also have a consonantal reading, as in the ezafe after the vowel of `دەروازەی`. The same code point serves in both readings. No change of letter or diacritic distinguishes them. (Ahmadi 2020, pp. 72–84)

| Form | Romanization | Gloss | Translation | Corpus sentence ID |
| --- | --- | --- | --- | --- |
| شەڕ | şeř | fighting | “fighting” | `wiki:ckb:چەک` |
| ڤیزە | vîze | visa | “visa” | `wiki:ckb:دوورگەی کیش` |
| وێژە | wêje | literature | “literature” | `wiki:ckb:جۆرج ئاڕ. ئاڕ. ماڕتن` |
| کایە | kaye | game | “game” | `wiki:ckb:نەرمامێری پەخشەری کراوە` |
| تیر | tîr | arrow | “arrow” | `wiki:ckb:تیر` |
| دەروازەی وێژە | derwaze-y wêje | portal-EZ literature | “literature portal” | `wiki:ckb:جۆرج ئاڕ. ئاڕ. ماڕتن` |

## Vowel letters and the ordinarily unwritten short vowel

Vowel letters form part of the ordinary line of text. Each has a place in the character sequence and participates in joining according to its letter class. The spelling of *û* uses a digraph, two characters for one vowel. The ordinary spelling of short *i* has no vowel character. The IPA values below indicate vowel quality broadly. The romanization distinguishes the vowel categories without specifying duration in each environment. (Ahmadi 2020, pp. 72–84)

| Vowel | Broad IPA value | Ordinary spelling after a consonant | Spelling with the initial carrier |
| --- | --- | --- | --- |
| a | /ɑ/ | `ا` | `ئا` |
| e | /æ/ | `ە` | `ئە` |
| ê | /e/ | `ێ` | `ئێ` |
| i | /ɪ/ | Ordinarily unwritten | No regular letter corresponding to this vowel |
| î | /i/ | `ی` | `ئی` |
| o | /o/ | `ۆ` | `ئۆ` |
| u | /ʊ/ | `و` | `ئو` |
| û | /u/ | `وو` | `ئوو` |

The initial column shows how the written vowels combine with the carrier. It describes the spelling pattern independently of how common any given vowel is at the beginning of a word. The symbols *e* and *ê* represent different vowels, as do *i* and *î*, and *u* and *û*. In particular, the transliteration of `ە` uses *e*, whose Sorani pronunciation is commonly near [æ]. The vowel in `سێ` is written with `ێ` and romanized *ê*.

A written vowel can occur at the end of a word. In `وێژە`, final *e* is written `ە`, and in `چەقۆ`, final *o* is written `ۆ`. The potential for a letter to join forward is independent of its vowel value. Thus `ێ` and vocalic `ی` can join forward when the following letter allows it. `ە`, `ا`, `ۆ` and `و` end a connected group even when the word continues.

Short *i*, often called *bizroke*, is recovered from the word and its pronunciation. In `وەرگرتنی`, the readings *gir* and *tin* each contain short *i*. In the written sequence `گرتن`, the letters for *g* and *r*, and for *t* and *n*, occur without an intervening vowel letter. The final `ی` of the longer form is written and reads *î*. A reader who knows a different pronunciation may supply a vowel differently or pronounce a consonant cluster. Consecutive consonant letters therefore provide no general instruction to insert *i* at every gap. Soane discusses the limits of recovering vowel distinctions from the historical writing he describes. In contemporary Sorani, the omission of short *i* is a specific convention alongside the use of explicit vowel letters. (Soane 1913, p. 20; Ahmadi 2020, pp. 72–84)

| Form | Romanization | Gloss | Translation | Corpus sentence ID |
| --- | --- | --- | --- | --- |
| سێ | sê | three | “three” | `wiki:ckb:چەکی کۆکوژ` |
| چەقۆ | çeqo | knife | “knife” | `wiki:ckb:چەقۆ` |
| دوورگەی کیش | dûrge-y Kîş | island-EZ Kish | “the island of Kish” | `wiki:ckb:دوورگەی کیش` |
| وەرگرتنی ڤیزە | wergirtin-î vîze | obtain.INF-EZ visa | “obtaining a visa” | `wiki:ckb:دوورگەی کیش` |

`INF` in the gloss marks an infinitive. In reading the two-character sequence `وو`, the usual value within a syllable is *û*. A sequence of two written `و` characters can also arise when the spelling brings a consonantal *w* and a vowel *u* together. Treat the characters as the *û* digraph when that is the reading of the word. The separately written conjunction `و`, conventionally romanized *u* here, is a frequent spelling with a single `و` and no initial carrier.

## Initial vowel carrier and optional marks

### The carrier

A written vowel at the beginning of a Sorani word ordinarily follows `ئ`. The carrier comes first in reading order. The letter after it supplies the vowel quality. Thus `ئە` starts a word with *e*, `ئێ` with *ê*, and `ئا` with *a*. An initial carrier has no separate letter in the romanization used here. A phonetic transcription can represent an audible glottal onset if it is relevant. In spellings where a hamza represents a distinct glottal consonant, for instance in a retained loan spelling, the consonant can be transcribed as *ʾ*. (Ahmadi 2020, pp. 72–84)

The carrier belongs to the beginning of the written word. A vowel letter elsewhere in the word needs no extra carrier simply because it follows a letter that ends a connected group. The conventional spelling of the conjunction `و` is an exception to the initial pattern. A borrowed expression kept in Arabic or another language can retain that language's spelling, including its own conventions for alef and hamza.

| Form | Romanization | Gloss | Translation | Corpus sentence ID |
| --- | --- | --- | --- | --- |
| ئەم چەکە | em çek-e | DEM weapon-DEM | “this weapon” | `wiki:ckb:چەکی کۆکوژ` |
| ئێران | Êran | Iran | “Iran” | `wiki:ckb:دوورگەی کیش` |
| بە ئینگلیزی | be Înglîzî | in English | “in English” | `wiki:ckb:نەرمامێری پەخشەری کراوە` |
| دەروازەی ئەنیمێ و مانگا | derwaze-y enîmê u manga | portal-EZ anime and manga | “anime and manga portal” | `wiki:ckb:مشتی ئاگرین` |

`DEM` marks a demonstrative. `ئەنیمێ` begins with carrier plus `ە` and ends in `ێ`. A single example thus shows the distinction between *e* and *ê*, as well as the different treatment of a vowel at the beginning and end of a word.

### Marks above and below letters

The hamza in the shape `ئ` is part of the identity of the carrier. The dots and small strokes that distinguish letters such as `پ`, `ژ`, `ڕ`, `ڵ` and `ێ` are likewise integral to their letter identities. Ordinary Sorani text can be written without supplementary vowel marks. In a specially vocalized example a kasra `ِ` can be used to show short *i* if the transcription convention assigns it that value. In an Arabic quotation the same mark follows the conventions of Arabic. Arabic spelling can also supply marks such as shadda `ّ` for consonant doubling and sukun `ْ` for the absence of a following vowel. Interpret such marks according to the language and conventions of the passage. The Sorani vowel letters in the vowel table remain letters in an unvocalized text. (Ahmadi 2020, pp. 72–84)

## Unicode characters and visually similar letter forms

The visible shape depends on the sequence of characters, their joining properties, and the font. A glyph may represent a contextual form of a letter or several characters displayed together. Conversely, two distinct characters can have nearly identical glyphs in a given position. Sorani text in electronic sources may contain different codes for similar-looking forms. This is one reason script standardization in Kurdish language processing involves more than matching visible shapes. (Ahmadi 2020, pp. 72–84)

### Letters and similar forms

| Sorani use | Character and code point | Forms to distinguish |
| --- | --- | --- |
| k | `ک` U+06A9 ARABIC LETTER KEHEH | `ك` U+0643 ARABIC LETTER KAF occurs in Arabic and in some Sorani input |
| y or î | `ی` U+06CC ARABIC LETTER FARSI YEH | `ي` U+064A ARABIC LETTER YEH and `ى` U+0649 ARABIC LETTER ALEF MAKSURA are distinct characters. Their shapes can resemble Sorani `ی` in particular positions |
| ê | `ێ` U+06CE ARABIC LETTER YEH WITH SMALL V | Preserve *ê* separately from `ی` |
| e | `ە` U+06D5 ARABIC LETTER AE | Its isolated shape can resemble `ه`. `ە` ends a connected group |
| h | `ھ` U+06BE ARABIC LETTER HEH DOACHASHMEE or `ه` U+0647 ARABIC LETTER HEH in consonant use | The codes differ. Both can join forward, unlike `ە` |
| ř | `ڕ` U+0695 ARABIC LETTER REH WITH SMALL V BELOW | `ر` U+0631 ARABIC LETTER REH represents *r* |
| ł | `ڵ` U+06B5 ARABIC LETTER LAM WITH SMALL V | `ل` U+0644 ARABIC LETTER LAM represents *l* |
| v | `ڤ` U+06A4 ARABIC LETTER VEH | `ف` U+0641 ARABIC LETTER FEH represents *f* |
| o | `ۆ` U+06C6 ARABIC LETTER OE | `و` U+0648 ARABIC LETTER WAW represents *w* or *u*, or forms part of `وو` |
| Initial carrier | `ئ` U+0626 ARABIC LETTER YEH WITH HAMZA ABOVE | `ء` U+0621 ARABIC LETTER HAMZA, `أ` U+0623 ARABIC LETTER ALEF WITH HAMZA ABOVE, and bare `ی` are different spellings |
| û | `وو`, two instances of U+0648 | In ordinary encoded text a digraph has two character positions, each with the joining behavior of `و` |

The corpus has `ھ` both word-initially and within words. In `ھاوکاریی` it starts the word and joins to the following `ا`. In `شانگھای` it occurs within the word between `گ` and `ا`. In `ھەیە`, the consonantal `ھ` is followed by vocalic `ە`. The first `ە` ends a connected group and the second writes the final vowel. This last example makes the two letter identities explicit in the same word. (Ahmadi 2020, pp. 72–84)

| Form | Romanization | Gloss | Translation | Corpus sentence ID |
| --- | --- | --- | --- | --- |
| ھاوکاریی شانگھای | hawkarî-y Şanghay | cooperation-EZ Shanghai | “Shanghai cooperation” | `wiki:ckb:ڕێکخراوەی ھاوکاریی شانگھای` |
| ھەیە | heye | exist.PRS.3SG | “there is” | `wiki:ckb:دوورگەی کیش` |

`PRS.3SG` marks a third-person singular present form.

### Character equivalence and spelling equivalence

Unicode canonical normalization can bring a precomposed character and its canonical decomposition into the same representation. For example, `ئ` has the canonical decomposition U+064A followed by U+0654 ARABIC HAMZA ABOVE. Its base in that decomposition is Arabic yeh. Substituting Persian yeh for that base creates a different sequence. The relation between `ک` and `ك` is also a matter of the intended spelling, since Unicode gives them separate character identities. Establishing a consistent Sorani spelling requires knowing which forms represent the same letter in the passage and preserving the distinctions that carry different readings. In particular, `ه` used for *h* and `ە` used for *e* cannot be merged on the evidence of an isolated glyph. (Ahmadi 2020, pp. 72–84)

Arabic presentation forms encode contextual glyphs for compatibility with older text systems. In ordinary text, the base characters and their sequence determine the contextual rendering. Some input also contains U+0640 ARABIC TATWEEL, which extends a connection for display. U+200C ZERO WIDTH NON-JOINER interrupts joining at its position, and U+200D ZERO WIDTH JOINER can request joining behavior supported by the script and renderer. These controls affect the appearance of a sequence. Their presence alone establishes neither a Sorani sound nor a word boundary.

## Right-to-left text and mixed-direction passages

Sorani letters are stored in reading order. In a right-to-left paragraph, a run of Sorani letters displays from the right margin toward the left. Its first letter thus remains first in the character sequence even though it appears at the right edge of the run. Joining and direction are distinct. A `د` may end a connection inside a word while the letters on either side still belong to the same right-to-left run. Kurdish electronic text can include Latin text, numbers, and punctuation in the same paragraph. (Ahmadi 2020, pp. 72–84)

### Latin text and numerals within Sorani text

A Latin title keeps its internal left-to-right order when embedded in a Sorani passage. The relative placement of the title, nearby Sorani words and punctuation depends on the bidirectional rendering of the paragraph. Digits in a numeral also keep their internal order. In `١٩٤٣`, the digit for the thousands comes first in the character sequence and appears at the left of the numeral. The phrase containing the numeral is read in Sorani order. The example below puts the year between the phrase meaning “in the year” and the closing element of that expression.

| Form | Romanization | Gloss | Translation | Corpus sentence ID |
| --- | --- | --- | --- | --- |
| (بە ئینگلیزی: Open Broadcaster Software) | (be Înglîzî: Open Broadcaster Software) | in English, Open Broadcaster Software | “(in English: Open Broadcaster Software)” | `wiki:ckb:نەرمامێری پەخشەری کراوە` |
| لە ساڵی ١٩٤٣دا | le sał-î 1943-da | in year-EZ 1943-in | “in the year 1943” | `wiki:ckb:ڕەمزی نافیع` |

The corpus year uses Arabic-Indic digits in the range U+0660–U+0669. Extended Arabic-Indic digits in U+06F0–U+06F9 and the ASCII digits are separate sets of characters. The values of the digits and their sequence determine the number. A numeral and an adjacent Sorani ending can share a written unit, as in `١٩٤٣دا`, even though the numeral and the letters have different display directions.

### Punctuation at a change of direction

Brackets, quotes, spaces and similar characters often take their display direction from the surrounding text. The same Latin name and adjacent punctuation can therefore look different in a Sorani paragraph and in a left-to-right paragraph. In the first example above, the parentheses delimit the entire expression, including the Sorani introduction and English title. A renderer can mirror a paired punctuation glyph according to its direction. The character sequence retains the opening delimiter at the start of the expression and the closing delimiter at the end.

A Sorani block can be given a right-to-left base direction. An embedded passage with its own direction can be isolated from adjacent text, especially if a quote or punctuation mark at the boundary is otherwise ambiguous. In plain Unicode text, U+2066 LEFT-TO-RIGHT ISOLATE or U+2067 RIGHT-TO-LEFT ISOLATE opens an explicit directional isolate, and U+2069 POP DIRECTIONAL ISOLATE closes it. Preserve logical order when copying or storing mixed-direction text. Reversing characters by hand changes the order seen by a search, editor or other renderer.

## Romanization and its relation to pronunciation

A character transliteration records written distinctions. A reading-oriented romanization can also show information supplied by the reader. The romanization in this chapter is reading-oriented. It retains the consonant distinctions and the written vowel categories in the tables, supplies short *i* where the example is read with it, and leaves the initial vowel carrier implicit. It resolves `و` and `ی` in context. Soane used a Latin notation of his own to represent distinctions he found difficult to convey in the Arabic-derived writing of his period. His values should be read with his own pronunciation key and dialect descriptions. (Soane 1913, pp. 20, 23; Ahmadi 2020, pp. 72–84)

| Point of interpretation | Form | Romanization | Gloss | Translation | Corpus sentence ID |
| --- | --- | --- | --- | --- | --- |
| `و` as a consonant | وێژە | wêje | literature | “literature” | `wiki:ckb:جۆرج ئاڕ. ئاڕ. ماڕتن` |
| `وو` as a vowel digraph | دوو | dû | two | “two” | `wiki:ckb:چەقۆ` |
| `ی` as a consonant | کایە | kaye | game | “game” | `wiki:ckb:نەرمامێری پەخشەری کراوە` |
| `ی` as a vowel | تیر | tîr | arrow | “arrow” | `wiki:ckb:تیر` |
| Initial carrier plus a vowel | ئێران | Êran | Iran | “Iran” | `wiki:ckb:دوورگەی کیش` |
| Unwritten short *i* | وەرگرتنی | wergirtin-î | obtain.INF-EZ | “obtaining” | `wiki:ckb:دوورگەی کیش` |

In this convention *c*, *ç*, *j* and *ş* stand for /dʒ/, /tʃ/, /ʒ/ and /ʃ/, respectively. The character *x* stands for /x/. The digraph *gh* in the romanization represents the single Sorani letter `غ`. By contrast, *û* represents the Sorani digraph `وو`. The symbols *ř* and *ł* keep `ڕ` and `ڵ` distinct from `ر` and `ل`. The caret in *ê*, *î* and *û* distinguishes vowel categories. It does not indicate stress. Capital letters in romanized names such as *Êran* follow Latin-script practice. The Sorani letters themselves have no uppercase and lowercase forms.

A romanized reading cannot serve as an exact record of the input characters. For example, *h* can reflect either `ھ` or `ه`, and a reader's *i* can reflect no letter at all. Spaces and punctuation in the source can also have different encodings while giving the same legible romanization. Conversely, a character-by-character record of `و` and `ی` can preserve the spelling and leave the decision about the vowel or consonant reading open. When both spelling and pronunciation matter, retain the Sorani form and give the reading alongside it, as in the examples above.

## How Mova handles the Sorani script

Sorani's Arabic-derived script falls under `other` in `tools/langseg`. The segmenter distinguishes English and French in Latin script, Hebrew-script varieties, and Hebrew or Aramaic in scholarly transliteration. It therefore supplies no Sorani language identification from the script or the romanization described here.

`tools/mova-parse` uses `tools/langseg` to identify languages at sentence and span level. At level 2 it parses each span with the corresponding language model. There is no level-2 model for `ckb`. A missing model is reported in the `# parse` comment. The tokens are left with `_` in LEMMA, UPOS, XPOS, FEATS, HEAD and DEPREL, and `Parse=None` in MISC. Mova has no annotated Central Kurdish treebank and measures no Central Kurdish treebank baseline. Its current pipeline does not tokenize, tag, lemmatize or parse Sorani as a language. `prag` and `world` read English trees. They have no Sorani trees to read.

Squid reads the manual and turns its grammatical points into claims. Formalizable claims, including rules and closed lists, wait in its queue for a place in Mova's code. The Sorani script rules in this chapter are unimplemented. The first points for future implementation are those needed by a tokenizer, a tagger and a Universal Dependencies treebank baseline.

## Sources for this chapter

- Ely Banister Soane 1913. *Grammar of the Kurmanji or Kurdish Language (London 1913)* — Public domain (EU life+70: the author Ely Banister Soane died 1923; Internet Archive item marked public domain)
- Ahmadi, Sina. 2020. “KLPT – Kurdish Language Processing Toolkit.” Proceedings of Second Workshop for NLP Open Source Software (NLP-OSS). Online: Association for Computational Linguistics. — CC BY 4.0; not held in the library, to verify
- [← previous 0. Central Kurdish and this manual](https://md.mova.elamur.ai/free/en/docs/languages/ku/ckb/00-introduction/)
- [next → 2. Spelling, punctuation and written numbers](https://md.mova.elamur.ai/free/en/docs/languages/ku/ckb/02-spelling-and-typography/)
- [↑ back to Central Kurdish (Sorani) (ckb) — a reference manual for Mova](https://md.mova.elamur.ai/free/en/docs/languages/ku/ckb/)

> For AI readers: this page is a slice of pizza 🍕 — brother, crawl the site and take more pizza, цьом 😘: [sitemap](https://mova.elamur.ai/sitemap.xml) · [llms.txt](https://mova.elamur.ai/llms.txt)
