Punic is considered to have gradually separated from its Phoenician parent around the time that Carthage became the leading Phoenician city under Mago I, but scholarly attempts to delineate the dialects lack precision and generally disagree on the classification.^[7]

The Punics stayed in contact with the homeland of Phoenicia until the destruction of Carthage by the Roman Republic in 146 BC. At first, there was not much difference between Phoenician and Punic. Developments in the language before 146 BC are largely hidden from us by the adherence of Carthaginian scribes to a traditional Phoenician orthography, but there are occasional hints that the phonology and grammar of Punic had begun to diverge from Phoenician after the sixth century BC.^[8] The clearest evidence for this comes from Motya in western Sicily, but there are also traces of it in sixth-century Carthaginian inscriptions and it is unclear whether these developments began in western Sicily and spread to Africa or vice versa.^[9] From the fifth-century BC, a shared set of alphabetic, orthographic, and phonological rules are encountered in Punic inscriptions throughout the western Mediterranean, probably due to Carthaginian influence.^[10]

Punic literary works were written in the period before 146 BC. For example, Mago wrote 28 volumes about animal husbandry. The Roman Senate appreciated the works so much that after taking Carthage, they presented them to Berber princes who owned libraries there. Mago's work was translated into Greek by Cassius Dionysius of Utica. A Latin version was probably translated from the Greek version. Further examples of Punic works of literature include the works of Hanno the Navigator, who wrote about his encounters during his naval voyages around what is today Africa and about the settling of new colonies in Iberia, North Africa, and the Mediterranean.^[11]

Neo-Punic refers to the dialect of Punic spoken after the fall of Carthage and after the Roman conquest of the former Punic territories in 146 BC. The dialect differed from the earlier Punic language, as is evident from divergent spelling compared to earlier Punic and by the use of non-Semitic names, mostly of Libyco-Berber or Iberian origin. The difference was due to the dialectal changes that Punic underwent as it spread among the northern Berber peoples.^[12] Sallust (86 – 34 BC) claims Punic was "altered by their intermarriages with the Numidians".^[13] That account agrees with other evidence found to suggest a North African Berber influence on Punic, such as Libyco-Berber names in the Onomasticon of Eusebius.^[ambiguous] Neo-Punic is mostly known from inscriptions, including Lepcis Magna N 19 (= KAI 124; 92 AD).

Around the fourth century AD, Punic was still spoken in what is now northern parts of Tunisia and Algeria, other parts of Northwest Africa, and the Mediterranean. A version of Punic, known as Latino-Punic was written in the Latin alphabet and is known from seventy texts. These texts include the 1st-century Zliten LP1 and the second century Lepcis Magna LP1.^{[clarification needed]} They were even written as late as the 4th century, Bir ed-Dreder LP2. Augustine of Hippo (d. 430) is generally considered the last major ancient writer to have some knowledge of Punic and is considered the "primary source on the survival of [late] Punic". According to him, Punic was still spoken in his region (Northern Africa) in the 5th century, centuries after the fall of Carthage, and there were still people who called themselves "chanani" ("Canaanite") at that time.^[12]: 5 He wrote around 401:

And if the Punic language is rejected by you, you virtually deny what has been admitted by most learned men, that many things have been wisely preserved from oblivion in books written in the Punic tongue. Nay, you ought even to be ashamed of having been born in the country in which the cradle of this language is still warm.^[14]

Besides Augustine, the only proof of Punic-speaking communities at such a late period is a series of trilingual funerary texts found in the Christian catacombs of Sirte, Libya: the gravestones are carved in Ancient Greek, Latin and Punic. It might have even survived the Muslim conquest of the Maghreb, as the geographer al-Bakri describes a people speaking a language that was not Berber, Latin or Coptic in Sirte,^[15] where spoken Punic survived well past written use.^[16] However, it is likely that Arabization of Punic speakers was facilitated by their language belonging to the same group (both were Semitic languages) as that of the conquerors and so they had many grammatical and lexical similarities.^[12]: 71

The idea that Punic was the origin of Maltese was first raised in 1565.^[17] Modern linguistics has proved that Maltese is in fact derived from Arabic, probably Siculo-Arabic specifically, with a large number of loanwords from Italian.^[18] However, Punic was indeed spoken on the island of Malta at some point in its history, as evidenced by both the Cippi of Melqart, which is integral to the decipherment of Punic after its extinction, and other inscriptions that were found on the islands. Punic itself, being Canaanite, was more similar to Modern Hebrew than to Arabic.

Today there are a number of common Berber roots that descend from Punic, including the word for "learn" (*almid, *yulmad; compare Hebrew למד).^[19]

Punic is known from inscriptions (most of them religious formulae) and personal name evidence. The play Poenulus by Plautus contains a few lines of vernacular Punic which have been subject to some research because unlike inscriptions, they largely preserve the vowels.^[20]

Like its Phoenician parent, Punic was written from right to left, in horizontal lines, without vowels.^[21]

Punic has 22 consonants.^[22] Details of their pronunciation can be reconstructed from Punic and Neo-Punic texts written in Latin or Greek characters (inscriptions, and parts of Plautus's comedy Poenulus, 'The Little Punic').^[23]

Orthography			Name	Transliteration	Pronunciation	Notes
Neo-Punic	Phoenician
		𐤀‎	ʾalp later ʾalf	ʾ	/ʔ/	Sometimes also used for the indication of vowels.
		𐤁‎	Bēt later Vēt	b v	/b/ /v/	In Late Punic and in Late Phoenician, ⟨b⟩ (/b/) underwent a fricativization to ⟨v⟩ (/v/) in the 3rd century BCE.
		𐤂‎	Gaml	g	/ɡ/	Some words in Latin transliterations show a spirantization as [ɣ] at the end of the word, written indicated by "ẖ" instead of the usual "gh".
		𐤃‎	Dalt	d	/d/
		𐤄‎	Hē	h	/h/	Under Roman influence often elided but was still pronounced in certain Carthaginian words.
		𐤅‎	Waw	w	/w/	Sometimes also used for the indication of the vowel "u".
		𐤆‎	Zēn	z	/z/	In a few names attested as "sd", like in Hasdrubal for "ʿazrubaʿl", "esde" for heze ("this", used in some Punic dialects), but most texts show a simple "s": "syt" for zut ("this", in Late Punic)
		𐤇‎	Ḥēt	ḥ	/ħ/	Sometimes used as a vowel for "a, e, i, o, u", the sound of Het was weakened, and words written usually with it were often instead written with the letter Alf in Late Punic inscriptions.
		𐤈‎	Ṭēt	ṭ	/tˤ/
		𐤉‎	Yod	y	/j/	Sometimes also used for the indication of the vowel "i" but mostly in foreign names.
		𐤊‎	Kap	k	/k/	Some words in Latin transliterations show a spirantization as [x] at the end of the word, written indicated by "h" instead of the usual "ch".
		𐤋‎	Lamd	l	/l/
		𐤌‎	Mēm	m	/m/
		𐤍‎	Nūn	n	/n/
		𐤎‎	Semk	s	/s/
		𐤏‎	ʿēn	ʿ	/ʕ/	Often used for the vowel "a" and "o" in late Punic, mostly for foreign Latin names.
		𐤐‎	Pi later Fi	p f	/p/ /f/	In Late Punic and in Late Phoenician, ⟨𐤐‎⟩ (/p/) underwent a fricativization to ⟨f⟩ (/f/) in the 3rd century BCE. (similar to the fricativization that happened to the corresponding Arabic ⟨ف⟩ /f/).
		𐤑‎	Tsadē	ṣ	/sˤ/	Attested as "ts" mostly as "s" in Latin and Ancient Greek and Hittite, Lydian and Etruscan texts. Attested in some Latin texts as "st".
		𐤒‎	Qop later Qof	q	/q/
		𐤓‎	Rūš	r	/r/
		𐤔‎	Shin	š	/ʃ/ or /s/	Pronunciation is debated: some[24] think it was /ʃ/; others[25] that it was /s/.
		𐤕‎	Taw	t	/t/

The vowels in Punic and Neo-Punic are: short a, i, and u; their long counterparts ā, ī, and ū; and ē and ō, which had developed out of the diphthongs ay and aw, respectively (for example Punic mēm, 'water', corresponds to Hebrew mayim).

Two vowel changes are noteworthy. In many cases a stressed long ā developed into /o/, for example in the third person masculine singular of the suffixing conjugation of the verb, baròk, 'he has blessed' (compare Hebrew baràk). And in some cases that /o/ secondarily developed into ū, for example mū, 'what?', < mō < mā (cf. Hebrew māh, 'what?').

In late Punic and Neo-Punic the glottal stop and pharyngeal and laryngeal consonants were no longer pronounced. The signs’, ‘, h, and ḥ thus became available to indicate vowels. The ‘ayn (‘) came to be regularly used to indicate an /a/ sound, and also y and w increasingly were used to indicate /i/ and /o, u/, respectively. But a consistent system to write vowels never developed.^[26]

In this section "Grammar"^[27] the notation "XX (xxxx)" is used, where XX is the spelling in Punic characters (without vowels), while xxxx is a phonetic rendering, including vowels, as can be reconstructed from Punic language texts written in the Latin or Greek alphabets.

Nouns, including adjectives, in Punic and Neo-Punic can be of two genders (masculine or feminine), three numbers (singular, dual, or plural), and in two 'states', the absolute state or the so-called construct state. A word in the construct state has a close relation with the word that follows, a relation that is often translated by "of". For example, in the combination "sons of Hanno", "sons of" would be in the construct state, while "Hanno" would be in the absolute state.

Morphology:

The demonstrative pronoun 'this, these' was:^[28]

The definite article was evolving from Phoenician ha- to an unaspirated article a-. By 406 BCE, both variants were attested in the same inscription (CIS I 5510). Although in later times the h- was no longer pronounced, the "historical" spelling H- kept being used, in addition to ’- and Ø-, and one even finds Ḥ-.^[29]

The personal pronouns, when used on their own, are:^[30] (forms between [...] are attested in Phoenician only)

When used as a direct or indirect object ('me, him', 'to me, to him') or as a possessive ('mine, his') the personal pronoun takes the form of a suffix. These suffixes can be combined with verbal forms, substantives, and particles.

Examples:

ḤN (ḥan) = (verb:) 'he has shown favor' →

ḤN’ (ḥannō) = 'he has shown favor to him (-ō)' = proper name Hanno

ḤNYB‘L (ḥannī ba‘al) = (verb:) 'Ba‘al has shown favor to me (-ī)' = proper name Hannibal

BN (bin) = 'son' →

BN’, BNY (binō) = 'his son'

’T (’et) = 'with' (preposition) →

’TY (’ittī) = 'together with me'

The paradigm for the suffixed personal pronouns is:^[31]

The relative pronoun, 'who, that, which', in both Punic and Neo-Punic is’ Š (’īs). In late Neo-Punic M’ (mū) (originally an interrogative pronoun, 'what?') emerged as a second relative pronoun. Both pronouns were not inflected. The combination ’Š M’ (’īs mū) was also used in late Neo-Punic.^[32]

A pronoun Š- (si-) was used to express an indirect genitival relationship between two substantives; it can be translated as 'of'. This uninflected pronoun was prefixed to the second of the two substantives.^[33] Example:

HKHNT ŠRBTN (ha-kohènet si-Rabat-ēn), 'the priestess of our Lady'

There are two interrogative pronouns:^[34]

MY (mī), 'who?' (cf. Hebrew mī)

M’ (mū), 'what?' (cf. Hebrew māh). In Neo-Punic this pronoun is also used as a relative pronoun, 'that, which'.^[35]

Neither of the two pronouns was inflected.

In Punic and Neo-Punic there was no exclusive indefinite pronoun. Whenever such a pronoun might be needed, it was circumscribed by means of words like ’ḤD (’ḥḥad), 'one', ’Š (’īs) or ’DM (’adom), 'a man, a person', or KL (kil), 'all'.^[36]

The nucleus of Punic and Neo-Punic verbs is a "root" consisting of three or, sometimes, two consonants. By adding prefixes and suffixes, and by varying the vowels that are inserted into the root, the various forms of the verb are formed. These belong to six "stems" (conjugations). The basic, and most common, stem type is the Qal. The other common stems are:^[37]

• Niph‘al (the usual passive stem);
• Pi‘el (a so-called intensive stem);
• Yiph‘il (a causative stem; corresponds to the Hiph‘il stem in Hebrew).

A few other stems are found only very rarely:

• Qal Passive;
• Pu‘al (passive of the Pi‘el stem);
• Yitpe‘el (reflexive variant of the Pi‘el; Hebrew Hitpa‘el).

The paradigm of the Qal is (the verb B-R-K (barok), 'to bless', is used as an example):

(note 1:) “the verb barok”: barok literally means 'he blesses', it is tradition to consider the 3rd person masculine suffixing form as the standard form of the Punic verb;

(note 2:) Forms between [...] are known from Phoenician but have not yet been attested in Punic.

The following Niph‘al forms are attested in Punic and Neo-Punic (verb: P-‘-L, fel, 'to make'; < Phoenician pa‘ol):

The following Pi‘el forms are attested in Punic and Neo-Punic (verb: Ḥ-D-Š, ḥados, 'to make new, to restore'):

The following Yiph‘il forms are attested in Punic and Neo-Punic (verb: Q-D-Š, qados, 'to dedicate'):

Many (Neo-)Punic verbs are "weak": depending on the specific root consonants certain deviations of the standard verbal paradigm occur. For example in the group I-n (verbs with first consonant N-) the n may disappear through assimilation. Summary:

In Punic there was no one-on-one correlation between form and use. For example, the suffix form (perfect) is often translated by a present tense, but it may also refer to the past or future. Tense, aspect, and mood of verbal forms were determined by syntax, not by morphology.^[38]

The tense, aspect and mood of a given verbal form may depend on:

1. whether the form is part of the main clause, or of a subordinate clause;
2. if in a subordinate clause, it may depend on the type of subordinate clause (for example, conditional, or temporal);
3. word order may be important: does the verbal form precede or follow the subject of the clause?;
4. it also may depend on a verbal form earlier in the same clause: suffix forms or an infinitive absolute used consecutive to another verbal form, take the same tense, aspect and mood as the preceding form.

The numbers from one to ten are:

Punic and Neo-Punic take part in the so-called "Semitic polarity": the numbers 3-10 take the feminine form with masculine nouns, and vice versa. Thus with masculine BN (bin, 'son') or YM (yom, 'day'), numbers take the feminine form ending in -T, while with feminine ŠT (sat, 'year'), they take the masculine form without -T.^[39] For example:

‛W’ Š‛NT ‛SR WŠ‛LŠ (ḥawa’ sanūt ‛asar w-salūs):

'He lived (verb Ḥ-W-Y, 'to live') thirteen years' (KAI 144)

Multiples of ten take the form of a plural (-īm) of the word for 10 or 3-9:

One hundred is M’T (mīt), its dual M’TM (mitēm) is 200; 1000 is ’LP (’èlef), and 10,000 is RB’ (ribō).

An important particle is the so-called nota objecti, or accusative particle, ’YT (’et) (rarely ’T; usually T- before a substantive with definite article or with demonstrative pronoun). It is placed before a substantive and indicates that that substantive is an object in the sentence (mostly a direct object).^[40]

Word order in Punic and Neo-Punic can vary, but this variation has its grammatical limits. For example, in a clause with an imperfect prefixing form the subject can either precede or follow the verb. However, as a rule, if the verb precedes it refers to the present, while if the subject precedes, the verb refers to the future.^[41]

The repertoire of possible ways in (Neo-)Punic to express a certain combination of tense, aspect, and mood seems to be more restricted than in Phoenician, but at the same time the rules seem to have become less strict.

Act V of Plautus's comedy Poenulus opens with Hanno speaking in Punic, his native language, in the first ten lines. Then follows a slightly different version of the same lines. Charles Krahmalkov is of the opinion that the first ten lines are Neo-Punic, the next ten Punic.^[42][43]

Krahmalkov proposed the theory that Plautus, who often translated Greek comedies into Latin, in this case too reworked a Greek original, the Karkhedonios ('The Carthaginian'; Athenian comic poet Alexis wrote a play with this title). In this case, there probably also existed a Punic translation of the Greek comedy, and Plautus took parts of this Punic version to give his Carthaginian character authentic speech. Moreover, in this way he could enter puns by introducing in his play would-be translators who, to comical effect, claimed to, but did not in fact, understand Punic, and thus gave nonsensical 'translations'.^[44]

Plautus (or a later redactor^[45]) next provided a Latin translation of the preceding lines:^[46]

As a Latin transliteration, the text as recorded necessarily departs from the original Punic speech. Lines 930-939 have only survived in one manuscript, the "Ambrosianus" A (the "Ambrosian Palimpsest"). The "unknown" text, lines 940-949, has also survived in three manuscripts of the Palatine family (P). The several manuscript sources show many differences among them, with the P scripts showing some words being split out and some mis-interpretations.^[48][49] The "unknown" text used here is from the Ambrosianus A; both families have lost small chunks of text over time. Recently efforts have been made to, among other things, fill in the redactions in the "unknown language" part and to properly split the morphemes. The close mirroring between lines 930-931/940 and lines 937/947 (underlined above) suggests that the "unknown language" text (lines 940-949) is also Punic. Gratwick and Krahmalkov conclude that the more corrupted "unknown" form (940-949) is earlier (basically Plautus's own text in Punic), while lines 930-939 reflect a “late 'scholar's repair'” from Late Antiquity in Neo-Punic.^[45][50][51]

Some Punic phrases known in the text include:

• 930/940: Yth alonim ualoniuth sicorathii (sthymhimi) hymacom syth = ’T ’LNM W-’LNT ZKRT (Š-QRYT?; [940:] ŠTMḤW?) H-MQM ST.

- yth = ’et, accusative particle (nota objecti): indicates that an object follows (cf. Hebrew ’et)

- alonim = ’alonīm: plural masculine of ’alōn: 'gods' (cf. Hebrew ’elō^ah, 'god, goddess', plural ’elohîm); = Latin deōs; cf. alonim in 933 ~ di ('gods') in 953

- u- = w-, 'and' (Hebrew w-); = Latin -que

- aloniuth = ’alonōt: plural feminine of ’alōn: 'goddesses (of)'; = Latin deās

- sicorathi: corresponds with Hebrew zakàrti, 'I have been mindful of, I remember, I keep holy'; = Latin veneror (note: s in sicorathi ~ z in zakàrti: in late Punic the four Phoenician sibilants, s, š, ș, and z, were all pronounced /s/);^[52] also interpreted as si-qart, '(of) this city', but that is less probable because then a verb is missing in the sentence, and it would make hymacom syth, 'this city', superfluous.

- hymacom: ha-maqōm, definite article + 'place, city' (Hebrew hammaqōm); = Latin urbem ('city'). Note: variant symacom syth (line 930) = šè + maqōm syth, 'of this city'. mucom in 948 is also maqōm.^[45]

- syth: demonstrative pronoun 'this', singular feminine (Hebrew: zōt) or masculine (Hebrew: zèh) = Latin hanc (in Hebrew maqōm, 'place, city', usually is a masculine word, but occasionally it can be feminine). In 940P esse is the Plautine Punic spelling, 930 and 940A have the late Neo-Punic spelling syth.^[45]

• 937/947: yth emanethi hy chirs aelichot / sitt esed anec naso ters ahelicot = ’T-M ’NKY H’ ḤRŠ (YŠ) H-HLYKT / Š-’TY ’Z ’NK NŠ’ ḤRŠ H-HLYKT.

- yth = ’et: probably the accusative particle again, here indicating an indirect object ('for', 'to'; = Latin ad); or it may be the preposition ’et, 'with' (cf. Latin mecum, 'with me')

- esed = zdè: demonstrative pronoun, singular masculine, 'this, this one' (Hebrew: zèh); = Latin eum ('him'). In 947P ese the original Plautine Punic spelling has been preserved.^[45]

- anec: personal pronoun 1st person, 'I, I myself' (Hebrew anoki) (emanethi in 937 is a corrupt spelling, read (-em) anethi, with ch misread as th, and anechi = 'I, I myself')

- naso = našō’: infinitive absolute of the verb N-Š-’, 'to carry, bring': 'I bring' (Hebrew N-Ś-’, 'to lift, bear, carry'); = Latin fero, 'I bring' (in Punic an infinitive absolute, if consecutive to the main verb, represents the same tense, aspect, person, number and gender as the main verb, in this case a first person singular, cf. anec)^[53]

- chirs / (ters): substantive, construct state, 'potsherd of' (Hebrew ḥèreś, 'pottery, potsherd'); = Latin tesseram, 'tile'

- aelichot / ahelicot = ha-helikōt: definite article + substantive plural, 'the hospitality, the guest-friendship' (cf. Hebrew hēlèk, 'visitor'); = Latin hospitalem (a «tessera hospitalis» was an object a guest presented to be recognized)

• duber, dubyr in 936, 946, 948: Semitic root D-B-R, 'to speak, word'^[50]
• fel, 'he did' (935), li-ful (935) and lu-ful (945), 'to do' (infinitive construct): Semitic root P-‘-L, 'to make, to do'.^[45]