I first got interested in عجميه when I found the wikipedia page for it. The sample images gave me some idea of how it worked, but I wanted to get a better feel for how it was actually used. So I spent a while looking through scanned manuscripts available online, and three in particular stood out to me for how well they could be understood by someone who only speaks modern Castillian (me). So I started putting together my own documents roughly describing the orthographical practices used by these manuscripts, and even tried writing some stuff of my own. And then some time ago I noticed that the Wikipedia page that got me into this in the first place had at different times been updated to have some really good info, and also to have some info that to me seems totally inaccurate. So I figure I might as well try writing down in my own website my own observations from an admittedly limited number of manuscripts. Later on I will also describe some changes I've done for my own personal style of writing, but those will be in a separate section because I don't want to present my own specific practices as historically accurate (they aren't always). To keep the explanations from being too long, I will assume some basic familiarity with the arabic script as used by modern Arabic,
Keep in mind that some letter shapes are different from those in MSA; particularly qaf is written with only one dot above (ڧ), fa is written with a dot below (ڢ), ḍammah is written as a curve without a loop. Shaddah often takes the form of an additional curve above fatħah and ḍammah or below kasrah (looking similar to the nuunated variants of these diacritics, but distinct because the additional line in nuunated fatħah and kasrah is not curved). When it's alongside fatħah, shaddah can take the omega-like form it does in most modern writing, but it is placed above fatħah instead of below it (resembling some modern font's shaddah+kasrah). Even when some of these alternate letterforms have their own unicode slots, I prefer using the standard variants in typing for the sake of compatibility: The most "nonstandard" usage on this page is Persian kaf, which is still common enough to be supported by most fonts (swash kaf ڪ looks more like the one used in Maghrebi manuscripts, but it exists in unicode specifically because it is distinct in Sindhi, and isn't supported in all fonts).
Overall Aljamía used the same letters and diacritics already found in Arabic to represent Iberian Romance phonology. Footnotes explain additional details about some letter pairs.
| Arabic Letter(s) | Transliteration | IPA Arabic | IPA Spanish (old→modern) | Spanish Equivalents (old→modern) |
|---|---|---|---|---|
| ب | b | /b/ | /b/ | b, v |
| ج | j | /dʒ/ | /(d)ʒ/→/x/ | j, g(e/i) |
| د | d | /d/ | /d~ð/ ¹ | d |
| ه | h | /h/ | /h/→/∅/ ² | h |
| و | w | /w/ | /w/ | u (See "Diphthongs") |
| ز | z | /z/ | /dz̻~z̻/→/θ/ | z → z, c(e/i) |
| ح | ħ (ḥ) | /ħ/ | ⁶ | |
| ط | ṭ | /tˤ/ | ⁶ | |
| ي | y | /j/ | /j~ʝ/ | y, i (See "Diphthongs") |
| ک | k | /k/ | /k/ | c, qu(e/i), k |
| ل | l | /l/ | /l/ | l |
| م | m | /m/ | /m/ | m |
| ن | n | /n/ | /n~ɱ~m/ ³ | n (m) |
| ص | ṣ | /sˤ/ | ⁶ | |
| ع | 3 (ʿ) | /ʕ/ | ⁶ | |
| ف | f | /f/ | /f/ | f |
| ض | ḍ | /dˤ/ | ⁶ | |
| ق | q | /q/ | ⁶ | |
| ر | r | /r/ | /ɾ/ ⁴ | r |
| س | ç (s) | /s/ | /(t)s̻/ → /θ/ | ç, c(e/i) → z, c(e/i) |
| ت | t | /t/ | /t/ | t |
| ث | ṫ (th) | /θ/ | ⁵ ⁶ | |
| خ | ẋ (kh) | /x/ | ⁵ ⁶ | |
| ذ | ḋ (dh) | /ð/ | /d~ð/ ¹ | d |
| ظ | ẓ | /ðˤ/ | ⁶ | |
| غ | g (ġ, gh) | /ɣ/ | /g~ɣ/ | g, gu(e/i) |
| ش | s (š, sh) | /ʃ/ | /s̺/ | s |
| ء | 2 (ʾ ) | /ʔ/ | ||
| اَ ، ـَ | a | /a/ | /a/ | a |
| اِ ، ـِ | i | /i/ | /i/ | i |
| اُ ، ـُ | u | /u/ | /o~u/ | o, u |
| ءَا ، ـَا | e (aa) | /e/ | e | |
| آ ، ـَا | aa (e) | /aː/ | ⁷ | |
| اِي ، ـِي | iy (ii) | /iː/ | ||
| اُو ، ـُو | uw (uu) | /uː/ | ||
| بّ | bb | /bː/ | /p/ | p |
| جّ | jj | /dʒː/ | /tʃ/ | ch |
| لّ | ll | /lː/ | /ʎ/ | ll |
| نّ | nn | /nː/ | /ɲ/ | ñ |
| رّ | rr | /rː/ | /r/ ⁴ | rr, r |
| شّ | ss | /sː/ | /ʃ/→/x/ | x → j (x) |
| اً ، ـً | ã (an) | /an/ | ⁷ | |
| اٍ ، ـٍ | ĩ (in) | /in/ | ⁷ | |
| اٌ ، ـٌ | ũ (un) | /un/ | ⁷ | |
| ـٰ | ā (aa) | /aː/ |
1: Allophonic in contemporary Iberian Romance and modern Castillian, but distinct in many Arabic varieties. Both used in Aljamía.
2. Ha used only for /h/ as part of the /f/ → /h/ → /∅/ sound shift. Earlier manuscripts still have fa. "H" which was already in Latin words not written.
3. Nun always used in syllable-final positions for Romance words, even before B/P. Mim only syllable-final in Arabic words.
4. Trilled R always has shaddah, even in positions where Latin orthography would only use single R.
5. Did not yet exist in old Castillian.
6. Used only in Arabic loanwords and Arabic texts.
7. Used only in Arabic texts.
Diphthongs: Rising diphthongs written using ya or waw with the appropriate vowel diacritic, with the preceding consonant vocalized with kasrah or ḍammah respectively (e.g. "siete" شِيَاتَا). Falling diphthongs are written either as word-initial vowels would be with alif (e.g. "veinte" بَااِنْتَا), or with ya or waw with sukuun (e.g. "ley" لَايْ) [Not sure if this depends on each author's style or if it varies by word within the same text]. Hiatuses don't seem to be distinguished, being written the same as if the same vowel sequence was a diphthong (e.g. "día" ذِيَ).
Consonant Clusters: "True" consonant clusters don't happen word-initially, and can only have two consonants mid-word; in these, the first consonant of the cluster is written with sukuun. Many words which would have clusters have a nearby vowel placed on the clustered consonants (e.g. "trabaxos (trabajos)" is not تْرَبَشُّشْ, but rather تَرَبَشُّشْ "tarabaxos"). Some clusters in modern Spanish had been lost in old Spanish and were deliberately reintroduced later; these were reduced both in Aljamía and in contemporary Latin orthography.
Shaddah: Its usage isn't always consistent. Sometimes it's found in consonants which weren't distinguished in contemporary or modern Romance, sometimes it's found in Arabic loans which don't originally have it, sometimes the Romance consonants distinguished by shaddah are written the opposite of how one would expect even in documents where the distinctions do mostly align with what's expected (particularly the distinctions between S/X J/CH can be inconsistent).
Long Vowels: Long i/u ometimes used in Romance words, but I'm not sure of the exact criteria. (As an example of both this and the above, Arabic مُسْلِمْ is in one instance loaned as مُّسْلِيمْ , getting a long vowel and shaddah which weren't originally present).
Particles: Sometimes attach to the following word. "Y" can be either alif+kasrah before a consonant or ya attached to the following word before a vowel (e.g. "siete y ocho y nueve" شِيَاتَا يُجُّ اِ نُوَابَا).
Punctuation: Only punctuation I've seen is ؞ used as a full stop, which doesn't seem to be used as such in any modern Arabic-script writing.
Loanwords: Words loaned from Arabic tend to be written with the same consonants they have in Arabic, but vocalized differently (presumably as they were pronounced by Romance speakers).
کُوَنْدُ بِنُ ءَالْ اُتُنُّ شَا مَرْجِّتَرُنْ اِ کَيَارُنْ لَشْ هُجَشْ، اُنْ ءَاشَّانْبُّلُ دَالْ فِرِيُ کَا بُّرُنْتُ اَبْرِيَ؞ شَا هِزِيَارُنْ مُجَّشْ کَلْسَشْ بَّرَ بُّرُتَاجَارْشَا دَا لَ تَانْبَّارَتُرَ رِّجِذَ؞ اُشَّ اَللَّهْ کَا بُّرُنْتُ لَّاغَا لَ بِّرِمَبَارَ؞
kuwandu binu el utunnu se marjjitarun i kayerun las hujas, un essenbbulu del firiyu ke bburuntu abriya. se hiziyerun mujjas kalças bbara bburutejerse de la tenbberatura rrijiḋa. ussa al·llah ke bburuntu lleġe la bbirimabera.
cuando vino el otoño se marchitaron y cayeron las (foja→)hojas, un (exemplo→)ejemplo del frío que pronto (habēre→)habría. se (fazer→)hicieron muchas (calça→)calzas para protegerse de la temperatura rígida. (lawšaa2a al·lah→oxalá→)ojalá que pronto llegue la primavera.
My overall goal is to keep most of how historical aljamía works intact, with the few additions being distinct enough to not be misinterpreted. In particular there are two things I try to address: 1. readability without diacritics 2. diphthongs and hiatuses. Also, some other tidbits not big enough to get their own section.
As far as I can tell, historical aljamía is reliant on vocalization diacritics in a way most modern languages written with the Arabic script are not. In the context of formal manuscripts such as the above, this isn't a big deal since those tend to be fully vocalized anyways. But in casual writing where diacritics aren't always used (and even if used, aren't easily readable with all fonts) most Arabic-script languages have long vowels (and sometimes short a) marked with consonants. So I like writing stressed syllables with stressed i/u with long vowels (e.g. "Quito" کِيتُ and "quitó" کِتُو ). Since alif is already being used for e this can't be done quite the same with long a, but there's other options. Arabic uses ة for final feminine -a, which overlaps with many instances of final a in Spanish, so since the f→h→∅ usage never happens syllable-finally I figured I'd use ه for final unstressed a. And since alif maqṣuurah as far as I can tell doesn't already show up in aljamía orthography outside Arabic words which have it (and final ya seems to always be dotted) I figured that'd be a good way to do final stressed a (e.g. "papa" بَّبَّه and "papá" بَّبَّى). In words loaned from Arabic I tend to prioritize keeping the original consonant sequence, so while under these criteria Alá would end in alif maqṣuurah I tend to write it with ha to stick closer to the original spelling. For one-syllable words I don't do a long vowel unless it's an open syllable, and even then sometimes I omit it for disambiguation purposes (e.g. "sí" شِي vs "si" شِ ).
This one in particular is inspired by Urdu's usage of ya hamzah for glides instead of glottal stops (which like Spanish, it does not have). I also like doing this because in the context of all other Arabic-script writing, it looks strange when there's sequences of two alifs in some aljamía falling diphthongs (at least I've never seen another instance of adjacent alifs that aren't at least like, separated by a space). So for hiatuses starting or ending with i/o~u I write them as if they were a diphthong, but with the hamzah variant of ya/waw instead (e.g. "día" دِئَه , "ahí" اَئِ , "bahía" بَئِئَه , "veo" بَاؤُ). For hiatuses involving a/e on both parts I put a on an isolated hamzah, which is enough to avoid adjacent alifs (e.g. "leer" لَاءَار , "lea" لَاءَ , "cae" کَءَا). Dunno how good these actually look to someone with more experience reading Arabic than I have, hopefully not terrible.
I generally don't use sukuun too often, and don't do the cluster vowel reduplication thing. For /ʃ~ʒ/ as a loaneme I use چ (e.g. "shopping" چُوبِّنغ), similar to Egyptian Arabic's usage of the same letter albeit for different reasons (ش already being taken). Because I have the relatively rare hierba/yerba distinction, I try to distinguish those in my writings; with most instances of ya with kasrah before it being /j/ for me and most instances where it doesn't have kasrah before it being /ʝ/, I try adding initial vowel i before word initial or intravocalic /j/ (e.g. "yerba" يَاربَه , "hierba" اِيَاربَا , "Ushuaia" اُشُوَاِيَه [I don't like using alif for the weak part of diphthongs but for some edge cases I don't have a better solution]). For loans from Arabic-script languages I try sticking as close as I can to the original spelling, but sometimes I do implement changes reflecting their pronunciation in modern spanish (e.g. عَجَمِيَة ← عَلجَمِئَه )
کُوَندُ بِينُ ءَال اُتُونُّ شَا مَرجِّتَرُن اِي کَيَارُن لَش هُوجَش، اُن ءَاشَّانبّلُ دَال فرِئُ کَا بّرُونتُ اَبرِئَه؞ شَا هِزِيَارُن مُوجَّش کَلسَش بَّرَه بّرُتَاجَارشَا دَا لَه تَانبَّارَتُورَه رِّيجِذَه؞ اُشَّ اَللَّه کَا بّرُونتُ لَّاغَا لَه بِّرِمَبَارَه؞
kuwandu biynu el utuwnnu se marjjitarun iy kayerun las huwjas, un essenbblu del fri2yu ke bburuwntu abri2yah. se hiziyerun muwjjas kalças bbarah bbrutejerse de lah tenbberatuwrah rriyjiḋah. ussa al·llah ke bbruwntu lleġe lah bbrimabera.
