I’ve already written about the djinn themselves, and mentioned that they have their own language. The djinnish language is something of a mix of Arabic and Sumerian with a syntax that, to my knowledge, is rather unique (but that’s for a later post).
Djinnish has a wide and unusual range of consonants. There are four full consonant “series” and several extras. The series are labial, alveolar, pharyngealized alveolar, and velar. The extras are palatals, uvulars, pharyngeal, and glottals. Consonant length is also phonemic. The following table summarizes this.
|p b pʼ ɓ
|t d tˤ dˤ
|k g kʼ ɠ
|s z sˤ zˤ θ̱ ɮ
Of particular note is [θ̱], a voiceless alveolar non-sibilant (and non-lateral) fricative. Also notable are the ejectives and implosives.
There is a phonological distinction between plain consonants and emphatic consonants, though not every consonant has an emphatic version. The following table summarizes these associations.
|Alveolar lateral glide
Although Djinnish has a large selection of vowels, its system of vowel harmony greatly restricts their use. There are three strong vowels: /u/, /e/, and /ä/. It is important to note that, between them, they cover three different heights, three different backnesses, and two different levels of rounding. Any combination of strong vowels can occur in a word, and in umlaut it’s always a strong vowel that drives the change. Lexical stress always occurs on a strong vowel; all lexical words have a strong vowel. (Some non-lexical particles and function words may lack a strong vowel; in these rare cases, non-harmonized weak vowels can occur.)
There are also four weak vowels, which harmonize with the stressed strong vowel in the word on either height or backness. The base forms of the weak vowels are /i/, /ɘ/, /o/, and /ɑ/. The first and third harmonize on backness, while the other two harmonize on height. However, when harmonizing with the strong vowel /u/, there are some exceptions. The following table summarizes the harmonization behaviour.
Strong vowels are always long (so for example /u/ is realized as [uː]), but weak vowels can be either long or short, and this distinction is phonemic. There are also diphthongs, which are formed from a strong vowel followed by a weak vowel.
Djinnish is written with a version of the Arabic alphabet, using a total of 43 letters. In addition to the Arabic orthography, linguists and dictionaries use a Roman transcription; there are at least three different attempts to standardize this, with minimal success so far. (I’ll use the second romanization for any future posts on the language.) The following table summarizes the mapping of the primary glyphs, in standard sorting order.
The weak vowels are written using various Arabic diacritics; the doubled diacritics are used for lengthened weak vowels. When placed on a vowel symbol, the diacritics form a diphthong. Unlike in Arabic, the weak vowel diacritics are rarely omitted (thus it’s functionally an abugida rather than an abjad). As in Arabic, lengthening of consonants is also phonemic; however, unlike Arabic, this is indicated with the sukun diacritic (eg long beh = بْ) rather than the shadda. (In transcriptions of digraphs, only the primary letter is doubled; the H or C remains singular.) This is not valid on the three vowel characters (ا ي و), which are always long. The following table summarizes the use of diacritics in Djinnish, together with a Roman transcription (the last line applies to any consonant phoneme).
|Arabic Diacritic (on beh)
Occasionally the vowel harmonies are made explicit in transcription. For that purpose, a superscript u, e, or a is used indicating the vowel with which it has harmonized. For example, 〈iᵘ〉 would represent [y], 〈oᵉ〉 would represent [ø], and 〈oᵃ〉 would represent [ɵ]. Since this is redundant information, it is usually only done in dictionaries except in occasional words where the vowel harmony rules are broken.
In the third roman transcription form, some consonants are written with digraphs. The lengthened forms of these consonants double only the primary letter, so for example “ppc” for /pʼː/. If there is ever a situation where a digraph needs to be disambiguated from a sequence of two consonants, the sequence form will insert a hyphen between the consonants, so for example “p-c” for /pɕ/. This hyphen should not be taken as an indication of syllable boundaries!