The alphabetic system was invented probably in the early second millennium bce for a local West Semitic language of the Syro-Palestinian area. Examples from that period are known since 1905; they have been found in Serabit el-Khadim, in Sinai, and have been called “Proto-Sinaitic” (to distinguish them from late Nabataean inscriptions called Sinaitic). Two graffiti similar to the Proto-Sinaitic texts have been identified in Wadi el-Hol, near Thebes, in Egypt. It is still debated where this system—deriving from the Egyptian one—was invented; it has been supposed, without satisfactory enough evidence, that it was created in Egypt for the needs of Semitic soldiers. Another, earlier hypothesis proposes that the system was invented in Sinai, where Egyptians used Semitic personnel to work in turquoise mines (Hamilton 2014, for a proposal of origin and development). The attested “alphabet” was based—according to the prevailing opinion—on the acrophonic system—that is, every graphic symbol was given the phonetic value of the first consonant of the Semitic name of each representation. In the Semitic languages, all words begin with a consonant, a feature that partially explains why vowels were not represented specifically. It was supposed by Ignace Gelb (1952), with good argument, that originally every writing symbol expressed a consonant plus a nonspecified vowel (consequently, he called the system a “simplified syllabary”).
The consonantal alphabet (often called abjad, from the order of the first letters in the Arabic system) was adapted to the cuneiform technique in Ugarit (North Syria) in around the fifteenth century bce, deriving from a model not yet identified. Probably toward the beginning of the Iron Age (around the twelfth century bce or slightly later), with the reopening of trade links, a schematized form of the consonantal alphabet, with a reduced number of signs according to the Phoenician phonological inventory, began to be commonly used. This schematized script is first attested on some arrowheads around 1200–1100 bce, and, especially, in a group of royal inscriptions from Byblos. Soon it spread to the Near East (adopted by the Israelites and their neighbors, the Moabites, Ammonites, Edomites, and the Aramaeans), the Aegean (adapted by the Greeks for their language), and the western Mediterranean (for the alphabet, see chapter 17, this volume, and references there).
According to this chronology and geographical distribution, it is possible to recognize different variants of the Phoenician script:
• Ancient Phoenician, until the end of the eighth century bce, attested in the eastern and western Mediterranean
• Phoenician proper, attested in the Levant until the first century bce (with developments)
• Punic, attested in the central and western Mediterranean from the end of the sixth century bce until 146 bce (conventional date, corresponding to the fall of Carthage to Rome)
• Neo-Punic: after 146 bce (conventional date)
The latest known inscription in the Phoenician alphabet, the Bitia inscription (Sardinia, KAI 173), dating to the second century ce (emperor Antoninus Pius, r. 138–161 ce), has no exact parallels and follows more the Phoenician tradition than the Punic one (it is generally, but wrongly, classified as neo-Punic).
Phases and Dialects
Owing to the lack of vowels in the writing system, the reconstruction of Phoenician is based on comparison with other Semitic languages, in particular Hebrew, and with Akkadian, Greek, and Latin transcriptions, particularly of personal names and sometimes place names. For the Late Punic period, the monologue and passages in Plautus’s Poenulus also help us reconstruct the language and its development (Sznycer 1967; Gratwick 1971). Thanks to the use of some consonants in neo-Punic inscriptions—mainly alef, he, ‘ain, and seldom ḥet, no more preserved in the late phonology of the language—used as vowels (matres lectionis), along with waw and yod, we can trace the language’s development and especially reconstruct its last phase.
According to chronology, Zellig S. Harris and Johannes Friedrich have divided Phoenician into the following phases:
East West
Ancient Phoenician (tenth–eighth centuries bce)
Middle Phoenician (eighth–fifth centuries bce) Punic (fifth–second centuries bce)
Late Phoenician (fifth–end first centuries bce) Neo-Punic (second century bce–second century ce)
This division is chronological and geographical, but only partially corresponds to epigraphically attested dialects. From this point of view, different proposals have been advanced. The dialectal classification proposed here is:
East West
Byblian: Ancient (around 1000–ninth century bce) and late (after gaps; late sixth–fourth centuries bce) Standard (from the late ninth until the middle/end of the sixth century bce)
Standard or Tyro-Sidonian (late ninth?first centuries bce, with gaps) Punic (from the late sixth century until 146 bce)
Late Punic (second century bce–second century ce).
The proposed division, and especially its chronology, aims to distinguish phases of the language but should not be interpreted strictly; particularly, the Punic and Late Punic phases overlap in space and time, reflecting sometimes a spoken versus written language (more conservative). The dialect of Byblos, archaic in respect to Standard Phoenician, and the Standard language (Tyro-Sidonian) are the only clearly identified dialects. Byblian underwent several phases, as follows:
1. Ancient Byblian, which in turn shows two phases: (a) the first or archaic one attested only by the Aḥīrōm inscription (KAI 1); and (b) the following ancient royal inscriptions (tenth–ninth centuries bce), which are more developed, particularly regarding the form of the pronominal suffix of the third-person masculine singular (Ø and -W, still -H in Aḥīrōm).
2. The phase of the Persian and Hellenistic periods, which has preserved some peculiarities of the ancient dialect (e.g., the suffix Ø and -W), while it adopted some features of the Standard dialect (relative ’Š; Z in the preceding phase).
A hypothetical North Phoenician dialect (Garbini 1988: 51–68) is not clearly identifiable. Finally, a Cypriot dialect, not well defined, was probably used.
The latest evidence of the Late Punic phase of the Phoenician language comes from about sixty inscriptions in Latin letters from Tripolitania, chronologically not well defined but probably lasting from the late first–second centuries until the fourth–fifth centuries ce; their language cannot, however, be classified as a “dialect,” as it is not sufficiently documented and the inscriptions in many cases are still obscure.
Phonology
Consonants
From the first Phoenician inscriptions documented, the language was composed of an inventory of twenty-two consonantal phonemes, corresponding to the twenty-two letters of the alphabet used:
’, b, g, d, h, w, z, ḥ, ṭ, y, k, l, m, n, s, ‘, p, ṣ, q, r, š, t
The phonemes can be classified, according to place and manner of articulation, as follows:
As compared with the reconstructed Common Semitic consonant inventory, and with what can be reconstructed for the second millennium bce (transcriptions in Egyptian, Amarna letters, Ugaritic), the Phoenician system was reduced. The reduction corresponded probably with the schematization of the alphabet (the Proto-Sinaitic texts have at least 27 letters). In respect to the Common Semitic system, the velar fricatives ḫ and ġ had merged with the pharyngeals ḥ and ‘ (for ex. *’aḫ > ’aḥ “brother”; *ġalmat > ‘almat “girl”); as the four interdental fricatives, ṯ had merged with š (*ṯalāṯ > šalōš), ḏ with z (ph. demonstrative z), θ and ẓ with ṣ (*θōr > ṣōr “rock,” *’arẓ > ’arṣ “land/country”). As in Ugaritic, ś was not distinct from š, whose pronunciation is debated. Double consonants (“geminated” or lengthened) are not marked in writing (except in some Late Punic examples).
Changing, dropping of consonants, assimilation, and dissimilation are attested during the history of the language. Alef tends to be dropped in double closed syllables—for example, RŠ for R’Š “head/cape”: Punic coins with the legend RŠ MLQRT; Phoenician ML’KT, “work” (KAI 10, 11), Punic MLKT (CIS I 3914,2).
In Punic, laryngeals and pharyngeals weakened quite early, alef first, (definite article written ’-) and tended to be lost in Late Punic (they were confused and dropped), as is shown by late orthographies; for example, B‘L ’MN for B‘L ḤMN (CIS I 960,3); ŠMḤ, “he heard,” for ŠM‘ (EHA 141,3); ‘WH, “he lived,” for ḤW’ (KAI 142, 2).
Vowels
Phoenician had originally the three vowels of Common Semitic: a, i, u and ā, ī, ū. However, the pronunciation of vowels underwent changes, as follows:
• Short vowels: i could change to e, and u to o.
• Long vowels: final ī could change to ē.
In the Canaanite languages (always), ā > ō (Canaanite shift)—for example, *maqām “place” > maqōm: macom in Poen. 940. In Phoenician, furthermore, there is a change á > ā > ō (not in Hebrew)—for example, *labán “white,” Phoen. λαβoν, “white” (Hebr. labān), in αβιβλαβον (Dioscorides, De materia medica 3.102, 3.122); ναδωρ “he vowed” (KAI 175, 3).
Vowels, as already noticed, were not represented in Phoenician orthography; some consonants to denote vowels (matres lectionis) were used rarely to represent foreign personal names—for example, HRN’ for Greek Ερηνη (KAI 56). A system for marking vowels with the series of the semiconsonants W and Y and the series of laryngeals and pharyngeals was used seldom in Punic (-’ for the morphemes -ō and -ā of the suffix of the third-person masculine and feminine), commonly in Late Punic, but systematically only in some centers, as in Tripolitania. Alef was used originally for any vowel, then specifically for o and e; he was used specifically for e (but also for any vowel, as, more seldom, ḥet); ‘ain was used frequently for a, yod for i, and waw for u.
From the orthography of late inscriptions, transcriptions in Classical authors, and the role of accent and reduction of diphthongs, it is possible to reconstruct a system consisting of a, e, i, o, u, short and long (the role of length in Late Punic is difficult to ascertain). The Canaanite shift *ā > ō, already attested in Amarna, remained operative, so that every secondary ā changed to ō (unlike in Hebrew); see earlier examples, and Ph *naš’ati > našōti “I carried” (with the supposed quiescence of ’; cf. Poen. 937 nasot). Later, ō changed to ū; for example, *ṯalāṯ > šalōš > šalūš “three” (salus in Sancti Aurelii Augustini Opera, sec. 4, pt. I, edited by I. Dijvak: Epistula ad Romanos incohata expositio, 13). On the other hand, there was a tendency for every a to shift to o in stressed syllables.
Reduction of vowels is attested in pretonic and propretonic syllables: pretonic i is elided in *dābirīm > dobrim “saying” (pl.) (Plautus, Poen. v. 935); pretonic a > a very short vowel of the kind of Hebrew shewa in *banāti > *benūtī, “my daughters” (bynythi, Plautus, Poen. v. 932).
Diphthongs
Diphthongs were normally reduced; for example, *bayt > bēt (written BT), “house,” *mawt > mōt (MT), “death” > Μουθ (god of death?) (Sanchouniathon in Philon of Byblos, FGrH/BNJ 790, F2 = Eusebios of Caesarea, Praep. Ev. 1.10.1).
Accents
Phoenician (as Hebrew) had usually a stress accent on the last syllable (final after the dropping of final short vowels), as these syllables were tone-lengthened when the final short vowels were dropped (Harris 1936: 25): *yatána > *yatán > yatōn; *nadára > *nadár > nadōr (ναδωρ).
Morphology
Nouns and Adjectives
Schemes
The traditional root used to denote the forms is qaṭal (“to kill”), usually written as qatal. For the conjugation (tenses and moods), we use here katab (“to write”; Gzella 2011).
When transcriptions are present, they show that Phoenician exhibits the same varieties of noun classes as the other Semitic languages. Most are triradicals, but there are also biradical and, seldom, uniradical or quadriradical nouns. Words always begin with a consonant. Vowels and prefixed or suffixed consonants specify the morphology of the word. For instance, from the root zbḥ, meaning “sacrifice,” “offer,” adding a prefix m-, it is possible to form the noun MZBḤ, “altar”; the prefix y- marks the verbal imperfective form YZBḤ, “he will sacrifice.” Nouns generally follow regular patterns in the position, length, and quality of consonants and vowels.
Biradical nouns are generally, as in the other Semitic languages, names of relationship: ’B (’ab), “father” and ’Ḥ (’aḥ), “brother,” as we see, for instance, in the transcriptions of names such as Abimilki, Aḫimilki, Αβιβαλος. The schemes attested are qal, qil, or qul (qāl, qīl, qūl), as YD (*yád > yōd), “hand”; GR (gir > ger), “client” (name Γεραστρατος); also long (qall, etc.; for example, KP [kapp], καππα) or with long vowel derived from a diphthong as BT (bēt < *bayt), “house.”
Triradical nouns may have one or two vowels. Attested patterns include the following:
• With one vowel (qatl, qitl, qutl); for example, ’LP (’alp), “ox”; MLK (milk), “king” (name Aβδιμιλκων); ŠRŠ (šurš), “root” (συρις). The transcriptions show that these nouns did not take a subsidiary vowel, as in the Hebrew “segolates.”
• With two vowels, short or long; for example, qatal, qatil, qatul; *qatāl > qatōl (DGN, Δαγων, name of a god, or “corn”); qatīl (NṢB [naṣīb], “stela”); qatūl (ḥanūn “gracious” in the name IḪa-(a)-nu-nu); *qātil > qōtel (ŠPṬ [šōpeṭ, lat. sufes]).
• With middle radical lengthened: qattīl (’DR [’addīr], “great,” “mighty” (toponym Rusaddir).
• A few nouns are reduplicated; for example, GLGL (gilgal), “wheel.”
Nouns with prefixes are common. The most common prefix is m-, indicating a place or an instrument; for example, MQDŠ (miqdōš), “sanctuary,” “holy place” (root qdš “to be holy”). There are a few examples with prefix t-, as TKLT (taklit?), “addition,” “complement” (root kly “to be complete”).
Nouns with suffixes denote mostly specific classes of adjectives; the most common is -y, forming adjectives, as in ordinals: ŠLŠY “third” < ŠLŠ (šalōš > šalūš), “three”; or in ethnics: ṢDNY (ṣidōniy), “Sidonian.” Other suffixes attested are -n and -t: ’LN (< ’L “god”); plural ’LM, then ’LNM “gods” (Poen. 940/930 alonim); R’ŠT (< R’Š “head”), “first fruits,” KAI 31 (re’šīt).
Gender, Number, and State
Nouns exhibit, according to their ending, a gender (masculine and feminine), a number (singular, plural; seldom dual), and a state (absolute and construct). Adjectives follow the same inflection.
Gender: Masculine nouns have no special ending, while feminine ones have the ending -át (> -ōt) or –t; for example, MLK (milk) “king”; f. MLKT (milkát> milkōt), “queen”; ‘LMT (‘almatt < *‘almant) I, “widow.” Names of lands, countries, and body parts are often feminine (as ’RṢ, “country”). Final -t drops in Late Punic, at least in the pronunciation.
Number: The markers of number are as follows:
• Masculine plural ends in -īm, written –m; for example, ’LNM, “gods” (Poen. 930/940 alonim).
• Feminine plural ends in -ōt > -ut; for example, ŠNT, “years” (šanōt > šanūt, transcribed in Latin as sanut).
• DLHT, “doors” (sing. dlt; cf. δελτα), is an extended plural, as in Hebrew ’mhwt, “mothers.”
• Dual, masculine, ends in *-aym > - ēm; feminine, ends in *-taym > -tēm; this morpheme appears rarely, in twin body parts, such as in YDM (transc. iadem KAI 178, 1), “hands”; M’TM “two hundreds,” KAI 76B 9 (mi’atēm?).
State: This is the specific form exhibited by names bound in a genitive relationship. A noun or adjective can be in absolute state or in construct state. It is in construct state when it is bound to a following complement in genitive (as in “the citizens of Byblos: B‘L GBL [ba‘lē Gubal]). In the other cases, it is in absolute state (unbound). The two words, bound and unbound, form a sort of chain, where the main stress falls on the second word. Consequently, the word in construct state has the following features:
• A vowel reduction (not seen in the orthography
).
• The -m of the plural ending is not present: -īm > -ī > -ē (e.g., MLK ṢR = milkē Ṣōr, “kings of Tyre”) (-īm > -ē) (the general explanation is that the ending -aym of the dual > -ēm > ‐ē was extended also to the plural).
• Cannot be determined by the article.
Determination
As in Hebrew (and the other Canaanite languages), determination is marked by a prefixed article h- (originating perhaps from a deictic element *han); it is prefixed and produces the gemination (lengthening) of the first consonant. This phenomenon (deduced from Hebrew) is proven by Late Punic orthographies, such as ‘MMQM (“the place”) for ham-maqōm (KAI 173, 5). As a result of the weakening of h, the article is written’- (already in Punic), seldom also ‘- or ḥ- in Late Punic.
The article is used more rarely and less regularly than in Hebrew. Not present in Aḥīrōm, it is attested for the first time in the inscription from Byblos KAI 4 (Yahīmilk), in the tenth century bce: KL MPLT HBTM ’L, “all the ruins of these houses/temples.” As in Hebrew, after proclitic prepositions, h- is elided. After determined nouns, attributive adjectives normally (but not always) receive the article; for example, H’LNM HQDŠM (KAI, 14, 19), “the holy gods.” With a determined noun, the demonstrative only seldom receives the article (the rule in Hebrew); for example, HSPR Z, “this inscription” (hassipr zī) (KAI 24, 15); HSMLM H’L, “these images/statues” (KAI 40, 3), is infrequent.
The Oxford Handbook of the Phoenician and Punic Mediterranean Page 30