Click to view table of contents

►II.21. HISTORICAL SOUND CHANGES

►II.21.A. THE CONSONANTS

►II.21.A.a. THE LABIAL STOPS b, f

§α. Arabic originally had two bilabial stops *p ف and *b ب . The voiceless bilabial stop changed to a voiceless labiodental fricative f. The labiodental nature of this sound is clear in Sibawayh’s description (§II.20.B.14.).

§β. The change *p > f in Arabic is not very ancient, because this change did not affect the Northwest Semitic languages (the closest relatives of Arabic, from which Arabic separated probably in the 3rd millennium BC). The change *p > f has affected loanwords from Akkadian or Sumerian. For example, the Akkadian parzillum “iron” appears in Arabic as firziluṋ فِرْزِلٌ “iron cutter” and farzala فَرْزَلَ “fettered [somebody].” The same word appears in Ugaritic as brðl 𐎁𐎗𐎏𐎍 “iron,” in Hebrew barzɛl בַּרְזֶל “iron,” and in Syriac parzlȃ ܦܰܪܙܠܐܴ “iron.”

The Arabic filizzuṋ فِلِزٌّ “mineral” seems to be another cognate of the same word.

———————————————————————————

۞ SIDE NOTE

The voiced b 𐎁 and ð 𐎏 in Ugaritic brðl 𐎁𐎗𐎏𐎍 seem to indicate that this word was not borrowed from Akkadian but perhaps from Hurrian. The ultimate origin must be the Sumerian AN.BAR = barzil /parᵗsil/ “iron.” The Sumerian consonant which is traditionally written b was probably a tenuis p, just like Hurrian p when it was not spirantized. For a Semitic-speaker after c. 2000 BC a completely plain p would have sounded more like the voiced b than the Semitic voiceless p, because the latter was not plain but aspirated.

———————————————————————————

§γ. The change *p > f is found also in the South Semitic languages and in neighboring African languages. The following is quoted from Wikipedia (it is without source but the writer sounds informed):

The stop [p] is missing from about 10% of languages that have a [b]. This is an areal feature of the “circum-Saharan zone” (Africa north of the equator, including the Arabian peninsula). It is not known how old this areal feature is, and whether it might be a recent phenomenon due to Arabic as a prestige language (Arabic lost its /p/ in prehistoric times), or whether Arabic was itself affected by a more ancient areal pattern. It is found in other areas as well; for example, in Europe, Proto-Celtic and Old Basque are both reconstructed as having [b] but no [p].

In Pliny’s Natural History ↗ (Book XII, chapter 32) an Arabian kind of incense is mentioned under the name carfiathum. Edward Lipiński (in Semitic Languages: Outline of a Comparative Grammar, p. 109) thinks that this means “autumnal” and he derives it from the Ancient South Arabian xrf “autumn” (c.f. Arabic xarȋfuṋ خَرِيْفٌ “autumn”). If this is correct, it shows the change *p > f in South Arabia already by the time of Pliny the Elder in the 1st century AD. Lipiński mentioned also the Sabaic word blṭ which designated coins widely used in ancient Yemen. He believes that this comes from Greek παλλάδες (palládes), a supposedly popular designation for the Athenian tetradrachms with the head of Pallas Athena (this etymology is not universally accepted, see Peter Stein (2010), in Coinage of the Caravan Kingdoms: Studies in Ancient Arabian Monetization, p. 318). If the etymology cited by Lipiński is correct, the Sabaic transcription blṭ shows that the Sabaic f was unsuitable for writing Greek π (p).

Perhaps the change *p > f was an areal feature that started in the Iron Age civilization of Yemen and spread from there to Ethiopia and northern Arabia.

§δ. According to M. C. A. Macdonald (in The Cambridge Encyclopedia of the World’s Ancient Languages), the Ṣafaitic letter f (Ṣafaitic is an Ancient North Arabian language of the Syrian Desert) was used to write Greek π (p) as well as φ (ph). For example, Greek Φίλιππος (Phílippos) was rendered in Ṣafaitic Flfṣ. The confusion between b and p, which is typical of modern Arabic, did not exist in Ṣafaitic. This indicates that Ṣafaitic f was not a real labiodental fricative. Perhaps it was an affricate ᵖf = p͡f (as in German Apfel) or a strongly aspirated pʰ.

The use of f ف (besides b ب ) to render p is known in early Classical Arabic. For example, fustuquṋ فُسْتُقٌ “pistachio” is from Middle Iranian pistag, fȃlȗðaǵuṋ فَاْلُوْذَجٌ “sweetmeat made of flour, water, and honey” is from Middle Persian pȃlȗdag, and firǵȃruṋ فِرْجَاْرٌ “pair of compasses” is from Middle Persian pargȃr (see ARABIC LANGUAGE ii. Iranian loanwords in Arabic, in Encyclopædia Iranica).

§ε. The last sound in Sibawayh’s list of the bad “supplementary Arabic letters” (§II.20.A.c.) is the “bȃʔ which is like fȃʔ” اَلْبَاْءُ اَلَّتِيْ كَاَلْفَاْءِ . This is perhaps the same or a close sound to the “letter which is between bȃʔ and fȃʔ” اَلْحَرْفُ اَلَّذِيْ بَيْنَ اَلْبَاءِ وَاَلْفَاءِ which Sibawayh mentioned in chapter 525:

وَيُبْدِلُونَ مِنْ الْحَرْفِ الَّذِيْ بَيْنَ الْبَاءِ وَالْفَاءِ‏ الفاءَ، نَحْو‏َ “الفِرِنْد” و”الفُنْدُق”‏.‏ وربما أبدلوا الباءَ لأنهما قريبتان جميعًا، قال بعضهم‏ “البِرِنْد‏”.‏ فالبَدَلُ مُطَّرِدٌ في كل حرف ليس من حروفهم، يبدل منه ما قرب منه من حروف الأعجمية‏. [كتاب سيبويه، باب اطراد الإبدال في الفارسية]

What is meant here is the sound p in foreign loanwords, which is changed in Arabic to f or b, e.g. in the word funduquṋ فُنْدُقٌ “hotel” which comes from Greek πανδοκεῖον “inn.”

Sibawayh did not say who were the Arabs who pronounced the “bȃʔ which is like fȃʔ“ and in what circumstances. Later authors could only speculate on what he meant. The linguist ʔIbn Yaʕȋš اِبْنُ يَعِيْشَ , who lived in Aleppo in the 12-13th centuries, says in his commentary onمُفَصَّلُ الْزَّمَخْشَرِيِّ  the following:

وَمِثَالُ الْبَاءِ كَالْفَاءِ قَوْلُهُمْ فِي بُورٍ “فُورٌ”، وَهِيَ كَثِيرَةٌ فِي لُغَةِ الْفُرْسِ. وَكَأَنَّ الَّذِينَ تَكَلَّمُوا بِهَذِهِ الْحُرُوفِ الْمُسْتَرْذَلِةِ قَوْمٌ مِنْ الْعَرَبِ خَالَطُوا الْعَجَمَ، فَتَكَلَّمُوا بِلُغَاتِهِمْ، فَاِعْرِفْهُ. [شرح المفصل لابن يعيش. فصل حروف العربية]

An example of the bȃʔ which is like fȃʔ is fȗruṋ [pȗruṋ] فُوْرٌ for bȗruṋ بُوْرٌ [“corrupt, hopeless”?]. This is common in the speech of Persians. It seems as if those Arabs who pronounced those unfavored letters were living with Persians and so they spoke like them. So know it.

Given the Ṣafaitic evidence mentioned above, and given that the change *p > f is apparently of Yemeni origin, it is not impossible that some Arabs in Sibawayh’s time preserved an older, harder pronunciation of f ف , and this could have been Sibawayh’s “bȃʔ which is like fȃʔ.”

►II.21.A.b. THE INTERDENTALS θ, ð, ð̣

§α. In many sedentary dialects of modern spoken Arabic the interdental sounds θ ث , ð ذ and ð̣ ظ are fortified to t, d, and ḍ. For example, the word θalȃθa·tuṋ ثَلَاْثَةٌ “three” is pronounced in Aleppo tlāte, the word ðahabuṋ ذَهَبٌ “gold” is pronounced dahab, and the word ð̣ahruṋ ظَهْرٌ “back” is pronounced ḍah(ə)r. However, in words borrowed from Standard Arabic the sounds θ , ð and ð̣ are changed to s , z and ẓ. For example, the word θabata ثَبَتَ “be(came) fixed; be(came) proven” is pronounced in Aleppo sabat, the word kaððaba كَذَّبَ “lied, said something untrue” is pronounced kazzab (the native cognate kaddab is used by some speakers), and ð̣ulmuṋ ظُلْمٌ “injustice” is pronounced ẓələm. Sometimes the same word is pronounced differently depending on the social context, e.g. the word ðakaruṋ ذَكَرٌ “male” is pronounced zakar in formal contexts and dakar in informal contexts, and ʔunθa·y أُنْثَىْ “female” is pronounced ʔunsa in formal contexts and ʔəntēye·t in informal contexts. The word ðiʔbuṋ ذِئْبٌ “wolf” is usually pronounced zəʔəb, but in personal names, proverbs, and certain old expressions it is pronounced dīb.

§β. The fortification of the interdentals is typical of sedentary dialects of southwestern Yemen (Aden), western Yemen (Ħudayda·t), western Arabia (Mecca), Greater Syria, Egypt, Sudan, and the Maghreb. This distribution corresponds to the historical movement of the Himyarites, but it is possible that this phenomenon was more widespread, given that the Ancient North Arabian (ANA) inscriptions from northern Ħiǵȃz (from Taymȃʔ, Dadȃn, and the Θamȗdic C inscriptions from Xašm Ṣanaʕ) and northern Naǵd (the Θamȗdic C inscriptions from ᵈŽildiyya·t mountain) used a character derived from the Ancient South Arabian (ASA) character s₃(which probably sounded ᵗs = t͡s) to write θ. This suggests that θ was pronounced ᵗθ by some groups in Ħiǵȃz and Naǵd c. 500 BC (see Ernst A. Knauf in The Qurʾȃn in Context (2010), volume 6, p. 215). The character θ was used in the ANA inscriptions from the Syrian Desert (known as Ṣafaitic), the Θamȗdic B inscriptions of Naǵd and the area between Madȃʔin Ṣȃliħ and Taymȃʔ, and the ASA of Yemen. The voiced ð was written in Taymȃʔ with the character z (c.f. Phoenician z ), but the rest of the ANA languages used various characters that (in my opinion) appear to be cognate with ASA ð (all of the Arabian characters for ð seem to be derived from the character z by the addition of a single stroke). Strangely, the ANA inscriptions from Ħismȃ (to the west of Tabȗk) used the ASA character ð to write θ, while the character θ was used to write g.

A Ṣafaitic inscription that was found on a stone at a cairn in the desert of Jordan (WH 1849) begins (as usual) by mentioning the name of the person who wrote it:

l-Whb-lh bn Θ̣n-ʔl bn Whb-lh bn Mlk

لوهبله بن ظنئل بن وهبله بن ملك

Luckily for us, this person knew how to write in Greek, and he left his name in Greek on an adjacent stone (WH 1860):

ΟΥΑΒΑΛΛΑΣ ΤΑΝΝΗΛΟΥ ΤΟΥ {Υ} ΟΥΑΒΑΛΛΟΥ

(Ouaballas Tannēlou [son] of {u} Ouaballou)

Ουαβαλλας (Ouaballas) corresponds to Classical Wahbu ʔ·Aḷḷāhi وَهْبُ اَلْلَّهِ “gift of God” with the Greek ending –ᾱς (–ās) of the masculine first declension. The spelling Ταννηλου (Tannēlou) for Θ̣n-ʔl ظنئل seems to indicate that the emphatic interdental θ̣ was not pronounced like a real interdental in the speech of this Ṣafaitic man. Ṣafaitic θ was rendered by Greek θ, e.g. the Ṣafaitic name Yθʕ was rendered Ιαιθεου (Iaitheou), but Greek θ also rendered Ṣafaitic t , e.g. the Ṣafaitic name Lbʔt was rendered Λοβαιαθου (Lobaiathou), and S₂mt was rendered Σαμεθος (Samethos) (for the reason why Greek θ and χ were used for writing Arabic t ت and k ك see §II.21.A.f.ζ.). Overall, it appears that the interdentals were fortified in Ṣafaitic, despite the conservative orthography which used the characters θ , ð and θ̣ . Since the interdentals were still distinguished in the orthography from the dentals, it is likely that they were not pronounced exactly like the dentals. They may have been pronounced ᵗθ, ᵈð and ᵗθ̣.

Confusion between t and θ is known in the ANA inscriptions from Dadȃn, where the word θalȃθuṋ ثَلَاْثٌ “three” was written θlt and tlt (Macdonald ibid.). According to Knauf (ibid. p. 221), the name of the Nabataean god Ðȗ ʔ·as₂-S₂ara·y ذُوْ اَلْشَّرَىْ  was written D-S₂ry in Ṣafaitic inscriptions from Ħawrȃn, but it was written Ð-S₂ry in inscriptions from Ħismȃ in northern Ħiǵȃz. Based on this he suggests that the interdentals were fortified to dentals in the speech of northern Nabataeans but not in the speech of the southern Nabataeans. However, the fact that Bedouins from Ħismȃ used ð in transcribing the Nabataean divine name does not necessarily mean that the Nabataeans themselves pronounced this sound. Since this is an Arabic word, an Arabic-speaking scribe can render it in his own pronunciation, especially if he believes his own pronunciation to be more correct. Besides, Macdonald (ibid.) says that d and ð were confused in inscriptions from Ħismȃ.

Given the evidence for the fortification of the interdentals in Ṣafaitic and Ħismaic (which are supposedly languages of Bedouins), one wonders whether any of the Ancient North Arabian languages preserved pure interdentals like the Classical Arabic ones.

In the Nabataean Aramaic alphabet (which is the ancestor of the Arabic alphabet) the characters t , d and ṭ were used for writing Classical Arabic θ, ð and ð̣. This is traditionally considered an Aramaism (see below §II.21.A.b.γ.). It is likely that the Nabataeans actually pronounced Classical Arabic θ, ð and ð̣ as t, d and ṭ. A direct evidence for this is found in the bilingual Ħarrȃn inscription of the Laǵȃ·t, which was written in 568 AD. In this inscription the Arabic name Ð̣ȃlimuṋ ظَاْلِمٌ was written طلمو in undotted Nabataean/Arabic and ΤΑΛΕΜȢ = ΤΑΛΕΜΟΥ (Talemou) in Greek. The Greek spelling shows that the descendants of the Nabataeans in the 6th century actually pronounced Classical Arabic ð̣ as ṭ (note that Greek τ ·t rendered Arabic ṭ, while θ (th) rendered t; see §II.21.A.f.ζ.). In the Petra papyri and Nessana papyri of the 6th-7th centuries AD, Arabic θ and ð were written with Greek θ and δ, which were also used for writing Arabic t and d. The Arabic words in these papyri belong to the speech of sedentary Arabs who inhabited Arabia Petraea (which is the same thing as the old Nabataea).

In cuneiform, ancient north Arabian θ was rendered t, e.g. the name of Yaθrib was written Ia-at-ri-bu = Yatribu in the Ħarrān stele of Nabonidus.

§γ. It seems that the fortification of the interdentals is an ancient development in Arabic. It is probably related to the similar development in Aramaic. In the Old Aramaic inscriptions the interdentals *θ, *ð and *θ̣ were initially written with the characters for s₁ ש, z ז and ṣ צ, but starting in the 8th century BC they began to be written with the characters for t ת, d ד and ṭ ט, which indicates the fortification *θ > t, *ð > d and *θ̣ > ṭ. The Old Aramaic inscriptions from Tall ʔ·al-Fəxēriyye·t تل الفخيرية used the character s₃ ס to write *θ before the general spelling with t ת was established. This should be compared to the use of the character s₃ to write *θ in Taymȃʔ and Dadȃn. That the Old Aramaic phonemes s₃, z and ṣ were pronounced as affricates *ᵗs, *ᵈz and *ᵗṣ can be inferred from the rendering of these sounds in Old Armenian. Old Armenian ᵗsʰeᵗsʰ ցեց “month” corresponds to Syriac s₃ȃs₃ȃ ܣܳܣܳܐ “month,” Old Armenian ᵈzētʰ ձէթ “olive (oil)” corresponds to Syriac zaytȃ ܙܰܝܬܐ “olive (oil),” and Old Armenian Naᵗsraᵗsʰi նածրացի “Nazarene, Christian” corresponds to Syriac Nȃṣrȃyȃ ܢܳܨܪܳܝܐ “Nazarene, Christian.” The Aramaic name of letter ṣ צ appears as ᵗsˀadey in early Georgian manuscripts. The Middle Iranian alphabet used the Aramaic letter ṣ צ to write ᵗš = t͡ʃ.

Perhaps the fortification of the interdentals was an areal phenomenon of the last millennium BC. It would seem that this started in the Aramaic-speaking world and spread from there to Arabia. It is now known that when Nabonidus conquered Ħiǵȃz in the 6th century BC, he established Aramaic as the official language there. The use of Official Aramaic in Ħiǵȃz continued during the reign of the local Liħyȃnids, at least in Taymȃʔ (M. C. A. Macdonald (2010), Ancient Arabia and the written word). These new discoveries show that the influence of Aramaic in Arabia was even older than the Nabataeans.

———————————————————————————

۞ SIDE NOTE

It would seem that some of the sound changes typical of Aramaic (which is an Iron Age language) have their origin in Late Bronze Age northern Syria. The fortification of the interdentals is first attested in Ugaritic (1300–1190 BC), where etymological ð was commonly written d (e.g. ðrʕ “arm” was also written drʕ; *ðqn “beard; old” was written dqn; *ʔxð “took” was written ʔxd). Interestingly, ð was confused in some roots with z (which stands for ᵈz), e.g. Ugaritic nðr “vowed” was also written nzr, zmr “sang” was also written ðmr, ðrʕ “arm” was also written zrʕ, and zrʕ “seed” was also written ðrʕ. The confusion between d, ð and z perhaps indicates that etymological ð was pronounced as an affricate *ᵈð. It appears that etymological θ (perhaps *ᵗθ in Ugaritic) was not confused with other phonemes. Perhaps the explanation for this asymmetry has to do with the aspiration quality of the voiceless sounds in Semitic. Since *ᵗθ would have been aspirated, it would have had a stronger fricative quality, and so it would have been more distinct phonetically than the unaspirated *ᵈð. A similar asymmetry is found in Old Akkadian, where etymological ð was not distinguished in writing from ᵈz, but etymological θ was distinguished from ᵗs (in Eblaite both θ and ð were distinguished from ᵗs and ᵈz).

An interesting phonetic development in Ugaritic is the occasional shift ð̣ > ɣ, e.g. Ugaritic ɣmʔ for *ð̣mʔ “be(came) thirsty” and nɣr for *nð̣r “paid attention; guarded.” This appears to be an antecedent of the distinctive Aramaic development *ṣ₂ > *ð̣ > *ʁ > ʕ, e.g. *ʔarṣ₂– > *ʔarð̣– > *ʔarʁ– (written <ʔrq> in Old Aramaic) > ʔarʕâ ܐܰܪܥܐ “land” (corresponds to Arabic ʔarḍuṋ أَرْضٌ “land”) (see §II.21.A.f.π.).

———————————————————————————

In my opinion, it is not impossible that the fortification of the interdentals existed in sedentary western Arabian dialects for many centuries before Islam. The fact that it was not mentioned by the early Muslim linguists does not necessarily mean that it did not exist. Those linguists failed to mention many linguistic features of the Arabic dialects of their days, probably because they did not think of those features as Arabic. The definition of the word “Arabic” for them was much narrower than the modern definition.

———————————————————————————

۞ SIDE NOTE

For the early Muslim linguists the word “Arabic” اَلْعَرَبِيَّةُ meant strictly the conservative Bedouinic dialects and certain sedentary dialects closely related to them. The sedentary dialects which were significantly different from the conservative Bedouinic dialects were excluded from Arabic and called by such names as Nabataean نَبَطِيَّةٌ (the word Nabataeans نَبَطٌ was applied both to Aramaic-speakers and Arabic-speakers), Jewish يَهُوْدِيَّةٌ (this word was applied to the Arabic spoken by the Arabian Jews, see Yahūd in the Encyclopaedia of Islam), and Ħimyaritic حِمْيَرِيَّةٌ (it is likely that this word was applied both to a Yemenite dialect of Arabic and to a non-Arabic Yemenite language; for the first see Chaim Rabin (1951), “Ancient West-Arabian” p. 42; for the second see Ḥimyaritic in the Encyclopedia of Arabic Language and Linguistics).

Some linguistic features that are typical of the modern spoken dialects of Arabic cannot be derived from the “Arabic” of the Classical linguists. It is probable that those features are inherited from ancient Arabic dialects that were not deemed Arabic by the early Muslim linguists, and thus they were completely ignored.

In my opinion, the fortification of the interdentals (which is nowadays very widespread, extending from Yemen to Morocco) is one of the ancient Arabic linguistic features that were completely ignored by the early linguists, because it did not exist in Bedouin speech (even today it does not exist in Bedouin speech). Probably the Bedouins thought of this feature as Nabataean, Jewish, Ħimyaritic, etc. It was not seen Arabic.

———————————————————————————

In modern Yemen the fortification of the interdentals is not widespread. It is found only in such places as Aden, Ħudayda·t, Soqotra, and in some cities of western Mahra·t. However, this patchy distribution is in itself an evidence for the antiquity of this linguistic feature in Yemen. This feature does not seem to be of South Semitic or Ancient South Arabian origin, rather it seems to be an areal feature that reached Yemen from the north by speakers of Arabic.

§δ. In Classical Arabic the sound θ ث was sometimes labialized to f ف . This is not a rare phenomenon in world languages (e.g. Russian Афины for “Athens”). The CA word θuwmuṋ ثُوْمٌ “garlic” has a variant fuwmuṋ فُوْمٌ . The adverb θumma ثُمَّ “next” has a variant fumma فُمَّ . The word θaɣruṋ ثَغْرٌ “opening, mouth” has a variant faɣruṋ فَغْرٌ . The word ǵadaθuṋ جَدَثٌ “grave” has a variant ǵadafuṋ جَدَفٌ . The word fattuṋ فَتٌّ “crack (in a rock)” has the variants θattuṋ ثَتٌّ and tattuṋ تَتٌّ . The verb ʔ·infaǵara اِنْفَجَرَ “burst out” has a variant ʔ·inθaǵara اِنْثَجَرَ . The verb farra فَرَّ “escaped” is perhaps cognate with θarra ثَرَّ “burst out.” The verb θȃra ثَاْرَ “erupted” is perhaps the origin of fȃra فَاْرَ “effervesced.” The verb daffa دَفَّ “pushed, hit” is perhaps cognate with daθθa دَثَّ “pushed, hit.”

The common word for “mouth” in CA is famuṋ فَمٌ . Another word for “mouth” is θummuṋ ثُمٌّ , which has a labialized variant fummuṋ فُمٌّ . (The word θumm ثُمّْ is used in many modern spoken dialects for “mouth”. In Aleppo it is pronounced təmm < *tumm.)

In modern sub-dialects of ʔAbyan and Ħaḍramawt the word θalȃθa·tuṋ ثَلَاْثَةٌ “three” is pronounced falȃf, and the word Daθȋna·tu دَثِيْنَةُ  is pronounced Dafȋna·t (Martine Vanhove, Yemen in the Encyclopedia of Arabic Language and Linguistics).

In the dialect of the rural Baħȃrina·t الْبَحَاْرِنَةُ of eastern Arabia there is a complete merger θ > f (Clive Holes, Bahraini Arabic in the Encyclopedia of Arabic Language and Linguistics).

PREVIOUS  NEXT