6 min read · 1,119 words
This is a personal effort to standardize the variety. I chose a neutral way for the spelling. I find this version the most efficient and regular.
If followed by "aa", they're always spelled ḍ, ṣ and ṭ rather than đ, s, t (ie pharygealized by default):
Seem [se̞ːm] ("to guess the price")
Ṣaam [sˤɐːm] ("to fast") | pharyngealization of "s" is implied upon preceding "aa"
Teeb [te̞ːb] ("to repent")
Ṭaab [tˤɐːb] ("for meal to finish cooking") | again pharyngealization is implied
Đeeq [ðe̞ːq] ("to taste")
Ḍaaq [ðˤɐːq] ("to grow narrower; to shrink") | and again
It kinda always goes like that, if pharyngealized then 'aa' otherwise 'ee'. Both vowels manifest a phonemic split.
Indicating pharyngealization is crucial:
Siif [siːf] ("sword")
Ṣiif [sˤiːf] ("summer")
Sabb [sæbː] ("to insult")
Ṣabb [sˤɐbː] ("to pour, to rain")
Traa [trɐː] ("you/she see/s")
Ṭraa [tˤrɐː] ("to soften")
Star [stɐr] ("to veil; to hide one's mistake/inconsistency")
Sṭar [stˤɐr] ("a line")
Đoll [ðʊlː] ("humiliation")
Ḍoll [ðˤʊlː] ("shadow")
Tunisian doesn't show consonant digraphs
'c' takes on /s/ before (e, i) and /k/ otherwise. Other consonants are all one-to-one.
Words spelled with "c" show they're loanwords. If they violate the language's phonotactics, they default back to "s" or "k":
Banc ("bench") -> Abnaac ("benches") [uuh idk but it doesn't look native]
Şleeke ("flip-flop") -> Şleeyec ("flip-flop's") [from Spanish 'chancla']
Cayyees ("road") -> Cayyeeceet ("roads") [from Spanish 'calles']
The use of 'y' for /iː/ as well as 'i' and 'u' as glides can be indicators for loanwords:
Clorophyl [klʊ.rʊ.'fiːl]
Espaanie [ɘs.'pɐː.njə]
Frombuaaz [frʊmb.'wɐːz]
Loanwords adapt in a systematic spelling
"e" (resp. "o") anywhere in the middle of a word makes /ɪ/ [ɘ] (resp. /ʊ/ [ʊ]) and at the end of the word /ə/ (resp. /u/)
"ee" (resp. "oo", "aa" and "ii") anywhere in the middle of a word makes /ɛː/ [e̞ː] (resp. /ɔː/ [o̞ː], /aː/ [ɐː] and /iː/ [iː]) and at the end of the word /ɛ/ [e̞] (resp. /ɔ/ [o̞], /a/ [ɐ] and /i/ [i])
The final (and any if you have the ɘ-ə merger) schwa has been chosen to be represent by 'e' because according to Consonant-stacking principle it becomes doubled when stacking a prefix. In all cases, it becomes ee (which corresponds to [e̞ː] pronunciation-wise)
varfe ['værfə] ("he verified")
varfeehom [vær'fe̞ːhʊm] ("he verified them")
djeeje ['dʒe̞ːʒə] ("hen")
djeejeet [dʒe̞ː'ʒe̞ːt] ("hens")
This system stays consisted in crucial examples. Compare 'me' ("not to be") whence 'meeni' ("I'm not but../I'm not") and 'ma' ("it was because") whence 'maani' ("It was because I...")
We define:
C a consonant cluster
V a vowel sound
Rule:
Stress is either ultimate or penultimate
The heavier syllable attracts stress
From heaviest to lightest:
CVVC
CVV
CVC
CV
If two syllables share the same characterization, that with the more amount of consonants wins
If they still draw, the penultimate wins
faaze : FAA-ze (CVV-CV: CVV) ['fɐː.zə] (fáaze)
faazto: FAAZ-to (CVVC-CV: CVVC) ['fɐːztu] (fáazto)
yetḍrab: yet-ḌRAB (CVC-C²VC: C²VC) [jɘt.'ðˤrɐb] (yetḍráb)
bayyett: bay-YETT (CVC-CVC²: CVC²) [bæj.'jɘt:] (bayyétt)
bayyet: BAY-yet (CVC-CVC: first) ['bæj.jɘt] (báyyet)
tbayyet: TBAY-yet(C²VC-CVC: C²VC) ['tbæj.jɘt] (tbáyyet)
dostour: dos-TOUR (CVC-CVVC: CVVC) [dʊs.'tuːr] (dostóur)
structuree: struc-tu-REE (CV-CVV: CVV) [strʏk.ty.'re̞] (structurée)
faazeet: faa-ZEET (CVV-CVVC: CVVC) [fɐː.'ze̞ːt]
structuur: struc-TUUR (CVC-CVVC: CVVC) [stryk.'tyːr ~ -ʏk] (structúur)
tenten: TEN-ten (CVC-CVC: first) ['tɘn.tɘn] (ténten)
franclee: FRANC-lee (CVC-CV: CVC) ['fraŋk.le̞] (fránclee)
This theory justifies spellings that feel odd, but they're valid:
'She thought for a moment' should be 'ꝁammmet' since the infinitive 'ꝁammem' is V1 (view) although the triple 'm' spelled looks like a typo, through hyphenation, stress holds and otherwise fails:
ꝁammmet: ꝀAMM-met (CVC²-CVC: CVC²) ['χɐmː.mɘt ~ 'χɐ.mə.mɘt]
if spelled ꝁa(m)memet, stress will be ultimate (ꝁa(m)-me-MET (CV-CVC:CVC)) which is wrong.
Suffixing or prefixing can irregularize stress; Tunisian doesn't particularly or necessarily have lexical stress (strees lexiic) , but certain cases of homography, might raise the need for it.
For example: toureŧ; compare ['tuːrɘθ] ("you/she inherit(s)") and [tuː'rɘθ] ("to be inherited"). While they both derive from the same verb 'wreŧ', the latter is a passive voice construction te- + wreŧ which happens to coincide with the former's conjugational one te- + wreŧ.
For such cases we may use non-standard accute diacritics (diacritiic es strees): tóureŧ vs. touréŧ
Another example: tékteb ("you/she write(s)) vs. tektéb ("to be written")
It actually happens that way for all CCeC verbs that admit a passive/reflexive
And yes, 'ew' -> 'ou'. So does 'ow' -> 'ou' and 'ey' -> 'ii' (aka: diphthong collapse)
Some allophony can be noticed with "a" in the presence of certain consonants.
Referring to the first column as æ-colors and the second as ɐ-colors.
If the consonant just preceding "a" is an ɐ-color, then "a" is phonemically transcribed [ɐ]:
Qarniit [qɐr.'niːtˤ] ("Octopus")
Raske ['rɐs.kə] ("to fare evade")
Ꝁamse ['χɐm.sə] ("five")
Ğaffel ['ʁɐf.fɘl] ("to fool")
Sṭal ['stˤɐl] ("bucket")
Tṣadmo ['tsˤɐad.mu] ("they were shocked")
Ḍalfe ['ðˤɐl.fə] ("cactus pad")
If the consonant just preceding is an æ-color then it would depend on the consonant cluster just after that same "a". If it contains at least one ɐ-color then it's phonemically [ɐ] otherwise [æ]:
Sabĝe ['sæb.ʕə] ("seven")
Barke ['bɐr.kə] ("a bless")
Now, when indicating an [æ] is crucial, it's then spelled with "ä", eg. bärke ['bær.kə] ("slacking; idling")
You'd never see a word ending in #Ce where C is an ɐ-color, instead they are spelled #Ca (meaning their theoretical schwas are compromised for [ɐ])
Tunisian now distinguishes three types of front vowels
Meet [me̞ːt]
Ĝaam [ʕæːm]
Ğaab [ʁɐːb]
They all historically come from the same phonemic representation of alif (ا). The divergence in articulation is documented as Imāla (or Imeele, you can see the phenomenon right there)
Pronunciation-wise, when you see the construction C₁C₂V and C₂ is a paired voiced consonant, C₁ is often voiced. Contrapositively, if C₁ is voiceless, C₂ is devoiced:
Sder ("chest") is phonemically /sdɪr/ but phonetically [zdɘr].
Other examples: tzeed ("to be increased") /tzɛːd/ ['dze̞ːd], sgat ("to fall; to catch") /sgat/ ['zgɐt] (regional), ꝁzar ("to look") /χzar/ ['ʁzɐr]
gtaates ("cats") /gtaːtɪs/ ['ktɐːtɘs] (regional), mteeĝhe ("her; hers") /mtɛːʕhə/ ['mte̞ːħhə ~ 'mte̞ːħːə]
In some accents, words ending in -eeCe are pronounced [-əːCə] instead of standard [-e̞ːCə].
Djeeje ['dʒəː.ʒə]
Hakkeeke [hæk.'kəː.kə]