►II.21.A.c. THE AFFRICATES AND SIBILANTS
The reconstruction of the Proto-Semitic (PS) sibilants has been difficult. The primary reason for this is that there are many sibilant phonemes in the Semitic languages, and it appears now that many of those were in PS affricates rather than sibilants.
Nowadays it has become common to reconstruct at least three affricate consonants for PS. These are *ᵗs (s₃), *ᵈz (z) ز , and *ᵗṣ (ṣ) ص . There is plenty of evidence supporting this reconstruction.
Many scholars have posited other affricates for PS. It has been suggested that the lateral sibilants *s₂ (ś) ش and *ṣ₂ (ṣ́) ض were originally lateral affricates *ᵗɬ and *ᵗɬ̣ (the emphatic lateral affricate *ᵗɬˀ exists in the Šeħri (ᵈŽibbȃli) language, and there is some evidence for its existence in ancient Gəʕəz). Moreover, it has been suggested that the interdental fricatives *θ ث , *ð ذ and *θ̣ ظ were originally affricates of some sort (usually they are said to have been *ᵗš, *ᵈž and *ᵗṣ̌).
These suppositions are supported by comparisons with other Afro-Asiatic languages. Personally I tend to believe the affricate hypotheses. These hypotheses make it easier to understand such alternations in Arabic as θamaruṋ ثَمَرٌ “fruit(s)” vs. tamruṋ تَمْرٌ “date(s) [of date palms];” θaʕlab– ثَعْلَبٌ “fox” vs. taʔlab– تَأْلَبٌ “ibex,” etc. The Proto-Semitic number *θamȃniy– ثَمَاْنِي* “eight” is derived from an original stem *θam–. The corresponding number in the Berber languages is tam “eight.” The Proto-Semitic root *rħṣ₂ رحض “wash” corresponds to Ancient Egyptian rḫt “wash clothes;” Arabic waṣȋfuṋ وَصِيْفٌ “servant” corresponds to Ancient Egyptian wdp /wṭp/ “servant;” Arabic wazana وَزَنَ “(he) weighed” corresponds to Ancient Egyptian wdn /wṭn/ “be heavy; weigh upon (someone);” Arabic ʔuðunuṋ أُذُنٌ “ear” corresponds to Ancient Egyptian *ỉdn /yṭn/ “ear.”
Examples of θ/t alternation from المزهر by السيوطي :
قال في الجمهرة: رجل كَنْتَح بالتاء والثاء جميعا: وهو الأحمق. والخَتْلة بالتاء والثاء: أسْفل البَطْن. وتُكْمَة بالتاء والثاء: اسم امرأة وهي بنت مُرّ أخت تميم ابن مُرّة. والكُتَّاب والكُثَّاب بالتاء والثاء: سَهْمٌ صغير يتعلم به الصِّبيان الرَّمْي. وتَخَّ العَجين والطِّين: كَثُرَ ماؤه ولانَ وقالوا: ثخَّ أيضا بالثاء والأولى أعلى. وفي أمالي ثعلب: الأكثم: الشعبان ويقال: أكتم بالتاء أيضا والمرأة كَثْماء. وفي فقه اللغة للثعالبي: يقال لمن نبتت أسنانه بعد السقوط مُثَّغِر بالتاء والثاء معا عن أبي عمرو. والهَتْهَتَة والهَثْهَثةُ بالتاء والثاء: حِكاية التواءِ اللسان عند الكلام. وفي المحكم: الثَّقْثَقة: الإسراع وقد حُكيت بتاءين. وفي المجمل: يقال لثأت به أمه: إذا ولدته سهلا وقد سمعته بالتاء أيضا. واستوتن المالُ: سمن وبالثاء أيضا. وفي المرصع لابن الأثير: يُقال للباطل ابن تُهْلَل وابن ثُهْلل. وفي تذكرة ابن مكتوم: التوي: المقِيم وبالثاء المثلثة أعرف. [السيوطي، المزهر في علوم اللغة وأنواعها، النوع السابع والثلاثون، معرفة ما ورد بوجهين بحيث يؤمن فيه التصحيف]
Examples of ð/d alternation from the same source:
قال أبو عبيد في الغريب المصنف في باب عقد له: خَرْدَلْت اللحم وخرذلتُه: قطعته. وادْرَعفَّت الإبل واذْرَعفَّت: مضت على وجوهها. واقدحر واقذحر. وما ذُقْتُ عَدُوفاً ولا عَذوفاً: أي مأكولا. ورجل مِدْل ومِذْل: وهو الخفي الشخص القليل اللحم. انتهى. وفي الإبدال لابن السكيت: الدَّحْدَاحُ والذَّحْذَاح: القصار الواحدة دَحْداحةٌ وذَحْذاحةٌ.
وفي الجمهرة: بَلْذَم الفرس: صَدْره ويقال بالدال أيضا. ودَحْمَلْتُ الشيء بالدال والذال والذالُ أعْلى: دَحْرَجْتُه على الأرض. ودفَفْتُ على الجريح بالدال والذال لغتان معروفتان والدالُّ الأصل: أجْهَزْتُ عليه. والخُنْدُع: الخسيس ويقال بالذال أيضا. وغَمَيْدَر: مُتَنَعِّم بالدال والذال. وقِنْدَحْر: وقِنْذَحْر: المتعرّضُ للناس. وحِرْدَوْن دابَّة أو سَبُع بالدال او الذال. وفي ديوان الأدب: مَرَد الخبز ومَرذَه: مَرَثَه. […] وفي الغريب المصنف عن أبي عمرو: أتتنا قاذِية من الناس وهم القليل وجمعها قواذ. قال أبو عبيد: والمحفوظُ عندنا بالدال. وقال أبو العباس الأحول: يقال للحمى أُمُّ مِلْذَم بالذال وقال غيره بالدال. قال علي بن سليمان الأخفش: ولست أنكر هذا ولا هذا. وفي فقه اللغة للثعالبي: الدَّألان بالدال والذال: مِشْيَةٌ في نشاط وخفة ومنها سُمِّي الذئب ذُؤَالة. وقال أبو عمرو الشيباني في نوادره: الذَّأَلان والدَّأَلان بالذال والدال. يقال: مرَّ يَذْأَل ويَدْأل في معنى واحد. وأجدعته وأجذعته: قطعت أنفه. وفي أمالي ثعلب: المُجَدَّع: المقطَّع الأنف والمجذَّعُ مثله. […] وفي كتاب الأيام والليالي للفراء: يقال مضى ذَُهْل من الليل ودَهْل بالذال والدال. وفي الصحاح: جدعته وأجدعته: سجنته وبالذال أيضا. وتمدحت خَواصِرُ الماشية: اتسعت شِبعاً بالدال والذال جميعا. ورجل مُنَجَّدٌ بالدال والذال جميعا أي مُجَرَّب. والمقذحر: المتهيىء للشر بالذال والدال جميعا.
ورجل هُدَرَة: ساقِط وهو بالدال في هذا الموضع أجود منه بالذال. وفي شرح المعلقات للنحاس يقال: جده يجُدّه: إذا قطَعه ويقال: جذه بالذال معجمة إذا قطعه أيضا. وفي شرح أدب الكاتب للزجاجي: الغذوي بالذال والدال معا. عن الليث: أن يباع البعير أو غيره بما يضرب هذا الفحلُ في عامه. وفي فقه اللغة: الخَرْدلة بالدال والذال: القَطْع قِطَعاً. وفي المقصور والممدود للقالي: الجادِل: الخَشِب الذي قد قَوي على بعض المَشْي وهو بالذال المعجمة قليل ويقال: جادل وجادن بالدال غير معجمة وهو الكثير الذي عليه أكثرُ العرب. وفي المجمل: جَذَف الرجل: أسرع بالدال والذال. والهيْدَبَى بالدال والذال: جِنْسٌ من مَشيِ الخيل.
(Although this is not usually acknowledged, it is possible that there is a relationship between PS *s₂alāθ–/*θalāθ– “three” and Proto-Indo-European *tréyes “three,” while PIE *kʷetwṓr “four” might be related to the Arabic root kθr كثر “many.”)
All the reconstructed PS phonemes except *ᵗs (and possibly *x̣) exist in Classical Arabic, but their sounds are not always the same as the reconstructed PS sounds. Thus, PS *ᵈz and *ᵗṣ became in Arabic simple sibilants z ز and ṣ ص .
►II.21.A.c.α. THE CENTRAL AFFRICATES *ᵗs, z, ṣ
§I. The PS phoneme *ᵗs corresponds to Ugaritic𐎒, Phoenician , Hebrew ס, Syriac ܣ , and Ancient South Arabian . In Arabic this phoneme was deaffricated prehistorically and it merged with s₁ س . It existed in such words as *ʔaᵗsara > ʔasara أَسَرَ “(he) took [someone] captive” (c.f. Hebrew ʔās₃ar אָסַר “(he) bound,” Syriac ʔes₃ar ܐܶܣܰܪ “(he) bound,” and Sabaic ʔs₃r “(he) took [someone] captive”) and *xaᵗsira > xasira خَسِرَ “(he) lost” (c.f. Hebrew ħās₃ēr חָסֵר “(he) be(came) empty, lacked,” Syriac ħəs₃ar ܚܣܰܪ “(he) wanted, lacked,” and Minaic xs₃r “(he) lost”).
§II. It is interesting that none of the Ancient North Arabian languages had a character for s₃. A character similar to the ASA was used in some ANA languages, especially sedentary ones, to transcribe θ rather than s₃ (see §II.21.A.b.β.). The absence of a character for s₃ means that none of the ANA languages still distinguished it from s₁. Thus, we may presume that the change *ᵗs > s₁ (and by analogy *ᵈz > z and *ᵗṣ > ṣ) happened already in Proto-Arabic (by which I mean the theoretical ancestor of the Arabic dialects, not the Ancient North Arabian dialects).
§III. Greek στ (st) is rendered in some early Arabic words by ṣ ص , e.g. Arabic ṣirȃṭuṋ صِرَاْطٌ “path” corresponds to Greek στρᾶτα < Latin strāta “paved road,” and qaṣruṋ قَصْرٌ “castle” corresponds to Greek κἀστρα “castle” < Latin castra “military camp.” Similarly, Middle Iranian ᵗš = t͡ʃ is rendered in some early Arabic words ṣ ص (besides š ش ), e.g. ʔ·Aṣ-Ṣȋnu اَلْصَّيْنُ “China” from Middle Iranian ᵗŠȋn–, ṣawlaǵȃnuṋ صَوْلَجَانٌ “scepter” from Middle Iranian ᵗšawgȃn–, ṣanǵuṋ صَنْجٌ “harp” from Middle Persian ᵗšang, ṣanȃruṋ صَنَاْرٌ “plane tree” from Middle Iranian ᵗšanȃr, ṣarmuṋ صَرْمٌ “hide, leather” from Middle Persian ᵗšarm, etc.
These words do not prove an affricate pronunciation of Arabic ص , because it is possible that they entered Arabic via Aramaic.
§IV. In the Nessana papyri of the 6th-7th centuries AD, the name of the town Nessana was written in Arabic نصان and in Greek Νεστάν- (Nestán-). Allegedly the name before the Muslim conquest appeared in Greek only as Νεσσαν- (Nessan-). The difference in spelling was taken as possible evidence that the Muslim conquerors introduced a new affricate pronunciation in the name (see Ahmad Al-Jallad (2014), Aṣ-ṣādu llatī ka-s-sīn – evidence for an affricated ṣād in Sibawayh?). However, this argument is unconvincing. The city of Aleppo was known before the Muslim conquest by the name Βέροια (Béroia), but after the conquest this name vanished and the city was known only as Ħalabu حَلَبُ . Should we conclude that the Muslims introduced the name Ħalab? This is certainly incorrect because this name is attested as early as the Yamħad (and possibly earlier). The Muslims probably revived a local name that was used by the non-Greek-speaking inhabitants of the city (that is, Syriac and Arabic speakers).
In a fragmentary Greek translation of sȗra·t CIII from the 9th century AD (Christian Høgel (2013), The Greek Qur’an: Scholarship and evaluations), the Arabic word ʔ·al-ʕaṣri اَلْعَصْرِ “the afternoon(gen.)” was transliterated ἀλέξαρ (aléxar). Ahmad Al-Jallad (ibid.) says that the use of ξ (x) in this word to render Arabic ص suggests an affricate realization of the latter. In my opinion, this is a bit far-fetched, especially given the date and nature of this document. More likely, the ξ in this word should be read /xs/ (voiceless velar fricative + s), and the writing έξα /exsa/ should be considered an attempt to render Arabic /ʕaṣ/. (In this same document the word ʔ·al-ʔuxdȗdi اَلْأُخْدُوْدِ “the trench(gen.)” (sȗra·t LXXXV:4) was transliterated ἄλαχουθ (álachouth), and the word ʔ·al-faǵri اَلْفَجْرِ “the dawn(gen.)” (sȗra·t LXXXIX:1) was transliterated ὄγερ (óger). The transliterations of this document are unreliable).
►II.21.A.c.β. THE LATERALS š, ḍ
§I. The Arabic š (s₂) ش corresponds to Hebrew שׂ and ASA . There is much evidence that this phoneme was originally a lateral consonant, probably a voiceless lateral fricative ɬ, and before that it was perhaps an affricate ᵗɬ (the sound ɬ is transcribed ś in many sources dealing with Semitic phonology).
The Arabic ḍ (ṣ₂) ض corresponds to ASA . This is believed to have been originally the emphatic counterpart of s₂, that is, it was also a lateral consonant (frequently transcribed ṣ́).
A lateral pronunciation of s₂ and ṣ₂ is found in the Modern South Arabian (MSAR) languages (which are not descended from the ASA). The MSAR s₂ is generally pronounced ɬ, and ṣ₂ is pronounced a weak ejective ɬˀ.
§II. Sibawayh described ḍȃd ض as an emphatic voiced lateral fricative ɮ̣. This pronunciation does not exist in any of the modern dialects. In fact, the modern dialects that (non-systematically) distinguish ض from ظ are found only in Yemen. In Ṣaʕda·t ض is pronounced ð̣ or a retroflex ᵗš, while ظ is pronounced ṭ (or ḍ?). In Laħiǵ ض is pronounced ð̣ and ظ is pronounced ḍ (Martine Vanhove, Yemen in the Encyclopedia of Arabic Language and Linguistics). Outside of Yemen (and in most Yemeni dialects) ض and ظ are not distinguished. The only modern dialects that have a lateral realization of ض/ظ are those of ʔAbyan and Γayl Ħabbȃn غيل حبان in Yemen. In those dialects both ض and ظ are pronounced (mainly) as a velarized ḷ (e.g. ʕaḷam for ʕað̣muṋ عَظْمٌ “bone(s)”). A variant of this pronunciation is ṛ (e.g. ʕaṛam “bone(s)”). In nearly all the other dialects ض and ظ merged as ð̣ or ḍ. In the sedentary dialects whose interdental sounds were fortified to dentals they merged as ḍ. In the Bedouin and Bedouin-influenced dialects they merged as ð̣.
In Modern Standard Arabic the pronunciation ð̣ is used for ظ and ḍ is used for ض , but this appears to be just an arbitrary standardization. It is not how these sounds were distinguished in any historical form of Arabic.
§III. The Classical Arabic word ʔ·al-qȃḍiy اَلْقَاْضِيْ “the judge” entered Spanish from Andalusian Arabic as alcalde “magistrate,” and the word ʔ·al-bayȃḍu اَلْبَيَاْضُ “whiteness” is the source of Spanish albayalde “white lead.” However, a Middle Arabic word from Andalusian Arabic was written in Roman characters nicayál/cayált and was said to mean “to spend summer.” This is apparently the word qȃyað̣a قَاْيَظَ “(he) be(came) in hot weather.” So it appears that in Andalusian Arabic both ض and ظ had a lateral pronunciation, like in some modern Yemeni dialects. In Malay there are words such as dloha “morning” from Arabic ḍuħaṋ ضُحًىْ “morning” and dla’if/la’if “weak” from ḍaʕȋfuṋ ضَعِيْفٌ “weak.” However, there is also dlalim/lalim “tyrannical” from ð̣ȃlimuṋ ظَاْلِمٌ “unjust,” which again shows that the donor dialect had merged ض and ظ . The Arabic words in Malay are probably of Ħaḍrami origin, and Andalusian Arabic is known to have been closely related to Himyaritic or Yemeni Arabic. Arabic loanwords with l for etymological ض are found also in East African and West African languages, also by influence from Yemen.
§IV. In the “Thompson prism” (Nineveh A) of Esarhaddon, which was written in c. 673 BC, the name of a north Arabian god was written ᵈRu-ul-da-a-a-ú, perhaps for *Ruɬ̣ȃw/*Ruɮ̣ȃw رُضَاْو* or *Ruɬ̣ȃyu/*Ruɮ̣ȃyu رُضَيُ/*رُضَاَيُ*. The variants Ruḍȃ رُضَاْ and Ruḍȃʔu رُضَاْءُ are known in early Islamic texts.
قَالَ ابْنُ إسْحَاقَ: وَكَانَتْ رُضَاءٌ بَيْتًا لِبَنِي رَبِيعَةَ بْنِ كَعْبِ بْنِ سَعْدِ بْنِ زَيْدِ مَنَاةَ بْنِ تَمِيمٍ، وَلَهَا يَقُولُ المُسْتَوْغِرُ بْنُ رَبِيعَةَ بْنِ كَعْبِ بْنِ سَعْدٍ حِينَ هَدَمَهَا فِي الْإِسْلَامِ “وَلَقَدْ شَدَدْتُ عَلَى رُضَاءٍ شَدَّةً … فَتَرَكْتُهَا قَفْرًا بِقَاعِ أَسْحَمَا” [سيرة ابن هشام]
It is believed that this is the same name as Ὀροτάλτ (Orotált) whom Herodotus mentioned ↗ in the 5th century BC as one of the two gods worshiped by the Arábioi of Egypt’s Eastern Desert. (N.B. the name Riḍwȃnuṋ رِضْوَاْنٌ which appears in some Islamic traditions as the name of an angel guarding Paradise has probably the same origin.)
In many of the Ancient North Arabian languages (Ṣafaitic, Θamȗdic B, Θamȗdic C, and Dȗmatic) the ASA character ð was adopted for writing etymological ṣ₂ ض . The character ṣ₂ was used in Dadȃn and Taymȃʔ, but it seems that in Taymȃʔ this character was modified by an additional stroke so that it became somewhat similar in appearance to ASA ð . For me, it is doubtful whether any of these languages preserved the original pronunciation of ṣ₂. According to Macdonald, Ṣafaitic ṣ₂ was always represented in Greek by σ (s), e.g. Ṣafaitic h-Ṣ₂fy was rendered in Greek Σαιφηνος (Saiphēnos) and Rṣ₂wt was rendered Ρασαουαθοϲ (Rasaouathos). The consistent use of σ to transcribe Ṣafaitic ṣ₂ (with no attested use of λ) shows that it was a sort of voiceless central fricative, probably *θ̣. This would explain why the ASA character ð was adopted for writing it. The discrepant fact that ð originally represents a voiced consonant should be explained in light of Sibawayh’s statement that ṭ ط is an emphatic version of the voiced d د , which was not disputed by any Arab linguist until modern times (vide §II.21.A.f.ε.). Since the voiceless emphatics are unaspirated, they are felt to be closer to the voiced plain counterparts, hence it is not surprising if the voiced ð was employed for transcribing the voiceless *θ̣.
The Nabataeans used the character ṣto write Arabic ṣ₂, which is continued today in the Arabic alphabet, where the character of ṣ ص is also used for writing ṣ₂ ض . Greek transcriptions of Arabic words that belonged to the Nabataeans and their later descendants usually used σ (s) to represent ṣ₂, e.g. the common Nabataean name Rṣ₂wt was transcribed Ρασαουαθοϲ (Rasaouathos). However, ζ (z) was also used sometimes, e.g. Ζαβεου (Zabeou) for Classical Ḍabʕuṋ ضَبْعٌ “hyena” (this personal name is rarely found nowadays, but it is attested over eighty times in Ṣafaitic inscriptions; see Al-Jallad, Ahmad, Robert Daniel, and Omar al-Ghul (2013), The Arabic toponyms and oikonyms in 17, in Ludwig Koenen, Maarit Kaimo, Jorma Kaimio, and Robert Daniel (eds.), The Petra Papyri II. Amman: American Center of Oriental Research, pp. 23-48.).
It appears that ṣ₂ and the emphatic interdental θ̣ (see §II.21.A.b.β.) were voiceless in Nabataea and the Syrian Desert (like in Aramaic and Hebrew). The voicing of these sounds is apparently of southern origin (since they are voiced in all the modern dialects, it seems that most of the Arabs pronounced them with voice already before Islam).
§V. In my opinion, many (if not all) of the sedentary ancient north Arabians dentalized the interdentals, while the lateral ṣ₂ was changed to an interdental θ̣/ð̣. This situation is found nowadays in some Yemeni dialects, where ض is pronounced ð̣ but ظ is pronounced ḍ (e.g. in Laħiǵ, vide supra §II.21.A.a.γ.II.2.). The same situation seems to have existed in the sedentary dialects of Greater Syria until fairly recently. The linguist اِبْنُ الْجَزَرِيِّ , who lived in Syria and Egypt in the 14-15th centuries AD, says in التمهيد في علم التجويد that “most of the [Greater] Syrians and some of the people of the Levant” (أَكْثَرُ الْشَّامِيِّينَ وَبَعْضُ أَهْلِ الْمَشْرِقِ) pronounced ض as ð̣ , but “most of the Egyptians and some of the people of the Maghreb” (أَكْثَرُ الْمِصْرِيِّينَ وَبَعْضُ أَهْلِ الْمَغْرِبِ) pronounced ض as ḍ. He does not say how ظ was pronounced in those dialects, but it was probably pronounced ḍ in all of them.
The antiquity of the pronunciations ð̣ ض and ḍ ظ may be inferred from this story mentioned by ʔAs-Suyȗṭiyy السيوطي in المزهر في علوم اللغة وأنواعها :
وَيُرْوَى أَنَّ رَجُلًا قَالَ لِعُمَرَ بْنِ الْخَطَّابِ “مَا تَقُولُ فِي رَجُلِ ظَحَّىْ بَضَبْيٍ؟” فعَجِبَ عُمَرُ ومَنْ حَضَرَهُ مِنْ قَوْلِهِ. فَقَالَ “يَا أَمِيرَ الْمُؤْمِنِينَ إِنَّهَا لِغَةٌ” وَكَسَرَ الْلَّامَ. فَكَانَ عَجَبُهُمْ مِنْ كَسْرِهِ لَامَ “لُغَةٍ” أَشَدَّ مِنْ عَجَبِهِمْ مِنْ قَلْبِ الْضَّادِ ظَاءً وَالْظَّاءِ ضَادًا.
قلت: هذا الأثر أخرجه القالي في أماليه، قال “حدثنا أبو عبد الله المقدمي حدثنا العباس بن محمد حدثنا ابن عائشة: حدثنا عبد الأعلى بن أبي عثمان الأسدي عن بعض رجاله قال: قال رجل لعمر “يا أمير المؤمنين، أَيُظَحَّىْ بِضَبْيٍ؟” قال “وَمَا عَلَيْكَ لَوْ قُلْتَ أَيُضَحَّىْ بَظَبْيٍ؟” قال “إِنَّهَا لِغَةٌ”. قال “انْقطع العتاب ولا يُضَحَّى بشيء من الوحشِ”.
It is told that a man said to ʕUmar bn ʔ·al-Xaṭṭȃb, “what do you say if a man gives an antelope as a [religious] sacrifice?” [ð̣aħħa·y bi-ḍabyiṋ ظَحَّىْ بِضَبْيٍ for ḍaħħa·y bi-ð̣abyiṋ ضَحَّىْ بِظَبْيٍ ] ʕUmar and those present with him were amazed from what he said. He said “O Commander of the Faithful, this is [my] language” [ʔinna-hȃ liɣa·tuṋ إِنَّهَاْ لِغَةٌ for ʔinna-hȃ luɣa·tuṋ إِنَّهَاْ لُغَةٌ ]; and their amazement from his saying “liɣa·tuṋ” for “luɣa·tuṋ” was even greater than their amazement from his saying “ð̣aħħa·y bi-ḍabyiṋ” for “ḍaħħa·y bi-ð̣abyiṋ.”
This anecdote (like most of the anecdotes ascribed to the early Muslim rulers) is probably not real history, but it serves to show the antiquity of two linguistic features that are typical of modern sedentary dialects, the first is the pronunciation ð̣ for ض and ḍ for ظ , and the second is the vowel i in place of Classical u.
§VI. Sibawayh’s description of the points of articulation (§II.20.B.) does not differentiate between a palatal and a post-alveolar points of articulation (§II.20.B.6.). He says that ǵȋm, šȋn, and yȃʔ are produced from the same point of articulation.
Sibawayh’s description of the points of articulation is generally precise. It is surprising that he would mix palatals with post-alveolars. There is at least one modern scholar (Edward Lipiński, Semitic Languages: Outline of a Comparative Grammar, p. 124) who has suggested that the šȋn described by Sibawayh is in fact a palatal fricative ç (like in German ich bin). There seems to be evidence for this in the 8th century AD bilingual Psalm fragment from Syria (see Joshua Blau (2002), A Handbook of Early Middle Arabic, p. 68). This is a fragment of the Septuagint with a translation in Middle Arabic (I.3.) written in Greek majuscules. The Greek Δ is used in the manuscript to transcribe Arabic د، ذ، ض، ظ ; Γ is used to transcribe ج، ع، غ ; Σ is used to transcribe س، ص ; Χ is used to transcribe ش، خ، ح ; Κ is used to transcribe ك، ق ; and T is used to transcribe ت، ط . It would be strange if Greek Χ (which sounded [x]) was used to transcribe the sound š = ʃ, but it would not be strange if it was used to transcribe the sound ç.
Lipiński says (ibid. p. 124) that in the modern Šeħri (ᵈŽibbȃli) language the former ɬ has evolved to ç, and he believes that a similar development happened in the Arabic described by Sibawayh.
§VII. It is probable that the pronunciation ɬ for ش was preserved in some early dialects of Classical Arabic. The following is quoted from the 13th century dictionary العباب الزاخر (under مضط):
[قال] الكسائي: الْمُضْطُ الْمُشْطُ. قال: يَجْعَلُونَ الشِّيْنَ ضَادًا بَيْنَ السِّينِ وَالضَّادِ، لَيْسَتْ بِضَادٍ صَحِيحَةٍ وَلَا شِينٍ صَحِيحَةٍ، وَهِيَ لُغَةٌ فِي رَبِيعَةَ وَالْيَمَنِ، يَقُولُونَ “اِضْتَرِ لِي” فِي مَعْنَى “اِشْتَرِ لِي” لفظًا ومعنًى.
[The linguist] ʔ·Al-Kisȃʔiyy [said], “ʔ·al-muḍṭu [ʔ·al-muɬṭu] is the same as ʔ·al-mušṭu [“the comb”]; [and he] said, “they make the šȋn into a ḍȃd that is between sȋn and ḍȃd, [it is] neither a correct ḍȃd nor a correct šȋn. It is used in Rabȋʕa·t and Yemen. They say ʔ·iḍtari-liy [ʔ·iɬtari-liy] in the meaning of ʔ·ištari-liy [“buy for me”].
ʔ·Al-Kisȃʔiyy was a contemporary of Sibawayh. He lived in Kȗfa·t in the 9th century AD. He says that šȋn had a sound “between sȋn and ḍȃd” in Rabȋʕa·t and Yemen. The word Rabȋʕa·tu رَبِيعَةُ was a general name for the east Arabian tribes. It is very nonspecific. Similarly, the word “Yemen” is very nonspecific since it can mean anywhere from ʔ·Aš-Šiħr to ʕAsȋr. However, it seems from what ʔ·Al-Kisȃʔiyy says that some people in eastern Arabia preserved the lateral sounding of šȋn.
This is another quote of ʔ·Al-Kisȃʔiyy in لسان العرب (under قشد):
[قال] الكسائي: يُقَالُ لِثُفْلِ الْسَّمْنِ الْقِلْدَةُ والْقِشْدَةُ وَالْكُدَادَةُ.
[The linguist] ʔ·Al-Kisȃʔiyy [said], “the sediment of butter is called ʔ·al-qilda·tu, ʔ·al-qišda·tu, and ʔ·al-kudȃda·tu.”
The alternation between š and l in pairs such as ʔ·al-qišda·tu/ʔ·al-qilda·tu is another evidence for the lateral pronunciation of šȋn. In fact, there are many Arabic pairs that show this alternation. There are also pairs that show the alternation ḍ/l and š/ḍ (see Lipiński’s book for some examples).
§VIII. One of the “supplementary Arabic letters” which Sibawayh considered good is the “šȋn which is like ǵȋm” اَلْشِّيْنُ اَلَّتِيْ كَاَلْجِيْمِ (see §II.20.A.b.). The explanation of this sound is perhaps found in chapter 569 ↗ of his book. Here is described a voiced allophone of šȋn that appeared in words such as ʔaʝdaqu for ʔaçdaqu أَشْدَقُ “good speaker.” Sibawayh characterized this allophony as ʕAarabiyyuṋ عَرَبِيٌّ “Arabic” (i.e. good Arabic) and kaθȋruṋ كَثِيْرٌ “numerous” (i.e. common).
►II.21.A.c.γ. THE SIBILANT s
§I. The only real sibilant that existed in PS may have been *s₁ س . It is unlikely that this phoneme was a plain *s. Very likely (almost certainly) it was palatalized *sʸ. This explains two strong historical tendencies of this phoneme:
- The tendency to become an interdental θ
- The tendency to become a post-alveolar š
Both of these tendencies may be explained by supposing an original *sʸ, which can easily become an alveolar θ̠ and a post-alveolar š. It is possible that PS *sʸ already had an allophone *θ̠ in the neighborhood of high vowels.
In the Ancient Egyptian language the 3rd person feminine singular suffix pronoun was written –s, but the masculine counterpart was written –f. This can be explained if we suppose that the masculine pronoun was pronounced *–sʸu > *–θu > *–fu. Some Egyptologists (Antonio Loprieno (1995), Ancient Egyptian: A linguistic introduction, p. 34) suggested that Ancient Egyptian s was pronounced sʸ.
The Biblical shibboleth story shows that some ancient Israelites pronounced s₁ as an alveolar *θ, while others pronounced it as a hard *s, or tried producing *θ by pronouncing *ᵗs (all the other interpretations of this story do not make sense).
It is likely that some Arabs pronounced s₁ as an alveolar *θ, given the following:
قَالَ أَبُو زَيْدٍ: مِنْ الْعَرَبِ مَنْ يَجْعَلُ الْسِّينَ تَاءً، وَأَنْشَدَ لِعلبَاءَ بْنِ أَرْقَمَ:
يَاْ قَبَّحَ اللَّهُ بَنِيْ السَّعْلَاةِ
عَمْرَو بْنَ يَرْبُوعٍ شِرَارَ الْنَّاتِ
لَيْسُوا أَعِفَّاءَ وَلَا أَكْيَاتِ
يُرِيدُ الْنَّاسَ وَالْأَكْيَاسَ [لسان العرب، سين]
According to an author quoted in لسان العرب , some Arabs would “turn sȋn into tȃʔ.” The given examples are by the poet علبَاءُ بْنُ أَرْقَمَ from Bakruṋ بَكْرٌ in eastern Arabia. The examples are the two pausal forms ʔ·an-nȃt for ʔ·an-nȃsi اَلْنَّاْسِ “the people(gen.)” and ʔakyȃt for ʔakyȃsiṋ أَكْيَاْسٍ “shrewd (gen. plur.).” The phonetic change s > t is hard to imagine, but the change θ > t is common in Arabic, and I have suggested above that this existed in Arabia long before Islam.
The fact that PS *θ merged with *s₁ in so many Semitic languages was no coincidence. These two phonemes had close sounds already in PS. The Old Akkadian language used the signs ŠA, ŠI and ŠU specifically to denote *θ, while *s₁ was denoted by the signs SÁ, SI, and SU₄. In Babylonian and Assyrian the old signs of *s₁ were abandoned. The signs ŠA, ŠI and ŠU became used to denote both *θ and *s₁, which indicates that these two phonemes had merged phonetically, most probably as an alveolar *θ. In the Ugaritic alphabetic writings the cuneiform š was rendered an interdental θ (except for words that have Ugaritic cognates with s₁). For example, the cuneiform divine name Aš-ra-tum was rendered in the Ugaritic alphabet ʔθrt, and the cuneiform divine name Ištar was rendered ʕθtr (this word seems to be the same as the Indo-European *h₂stḗr “star”). In my opinion, the Ugaritic Kθr w-Xss /*Kōθaru wa-Xasȋsu/, which is a title of Enki, the Sumerian god of terrestrial water, is perhaps cognate with the Sumerian divine name KI.ŠÁR “All-land” (the Arabic Kawθaru كَوْثَرُ , which is defined as the name of a “river in Paradise,” has probably the same origin).
The behavior of the phoneme <š> in Babylonian and Assyrian (e.g. the easy change š > l) indicates that it was not a hard s. It probably was an alveolar θ. Later (in the Iron Age?) this was retracted to a true š in Babylonian (like in Hebrew, Aramaic, and partly in the Modern South Arabian languages), but in Assyrian it was fronted to a hard s (like in Arabic, Ethiopic, and probably Sabaic).
§II. M. C. A. Macdonald suggested that Ancient North Arabian s₁ was pronounced “like something approaching š.” Although I do not think that he is mistaken in principal, the evidence which he based himself on is pretty weak. He referred to a claim that Sibawayh described the pronunciation of sȋn as š, and he cited the Dadȃnitic and Ṣafaitic transcription Bʕl-S₁mn for the Aramaic name of the great Syrian sky-god, Baʕl-S₁amȋn “lord of heaven.” If the Aramaic s₁ sounded š, it is not necessary that the Arabic counterpart should have had the same sound (any Arabic-speaker with some familiarity with Aramaic can tell that the Aramaic word šamȋn is cognate with the Arabic word for “heaven,” and so he/she may modify the pronunciation accordingly). According to Macdonald, Ṣafaitic s₁ transcribed Greek σ and Latin s in transcriptions such as Mrṭs₁ for Μύρτος (Mýrtos), Tts₁ for Titus, and Grgs₁ bn ʔqlds₁ for “Γεώργιος (Geṓrgios) son of Κλαύδιος (Klaúdios)” or “Georgius son of Claudius.” These transcriptions show that Ṣafaitic s₁ was not very close to š.
In my opinion, it cannot be dismissed that some Arabs participated in the Northwest Semitic shift s₁ > š. In Aleppo Arabic there are a few words which have š for etymological s₁. For example, the root s₁ṭħ “flat” is pronounced in Aleppo šṭħ (thus, in Aleppo the word s₁aṭṭaħa سَطَّحَ “laid flat (tr.)” is pronounced šaṭṭaħ). Another possible example is the word nəšəħ “dirt.” This word does not seem to be derived from the root ns₂ħ but rather from ns₁ħ “dirt” (whence comes the word mins₁ȃħuṋ مِنْسَاْحٌ “sweeper, something used from sweeping dirt,” in the Lisȃn الْمِنْسَاحُ شَيْءٌ يُدْفَعُ بِهِ الْتُّرَابُ وَيُذْرَى بِهِ ). In some dialects of western Syria the root s₁qṭ “fall” is pronounced šqṭ or qšṭ. People will often say that these examples are part of the Aramaic substratum in Syrian Arabic, which cannot be dismissed, but it is also possible that some of the early northern Arabs participated in the shift s₁ > š.
§III. In Arabic there are pairs that show an alternation between θ and s. For example, θbt “be(come) fixed” vs. sbt “sleep,” both of which are related to byt “dwell, reside; house;” skn “calm, feel safe” vs. θkn “grave, barracks,” both of which are related to knn “calm; protection;” sȃħa سَاْحَ “spread” vs. *θȃħa > fȃħa فَاْحَ “spread;” saṭaħa سَطَحَ “flattened” vs. *θaṭaħa > faṭaħa فَطَحَ “flattened;” saṭama سَطَمَ “cut” vs. *θaṭama > faṭama فَطَمَ “cut;” and *θalakun > falakuṋ فَلَكٌ “orbit” vs. salaka سَلَكَ “went in way.” An alternation between θ and š is perhaps seen in šaṭara شَطَرَ “divided” vs. *θaṭara > faṭara فَطَرَ “divided,” and šaqqa شَقَّ “tore open” vs. *θaqqa > faqqa فَقَّ “tore open.” The verb fariħa فَرِحَ “be(came) happy” might be related to the root šrħ. An alternation between ð, š, and ṣ is seen in ðhb “gold,” šhb “white; meteor,” and ṣhb “yellow” (c.f. hbb “glow of star,” in the Lisȃn وَهَبَّ اَلْنَّجْمُ: طَلَعَ ).
قال ابنُ السكيت في الإبدال: يقال: أتيتُه مَلْس الظَّلام ومَلْث الظلام: أي اخْتِلاط الظلام. والوَطْس والوَطْث: الضَّرْب الشديد بالخُفِّ. وناقة فاسِج وفاثِج وهي الفتيَّة الحامل. وفُوهُ يجري سَعَابيب وثعابيب وهو أن يجري منه ماء صاف تمدد.
وسَاخَتْ رِجلهُ في الأرض وثاخَت إذا دخلت. وفي الجمهرة: يقال جىء به من حيثك وحَيْسِك: أي من حيث كان. وفي ديوان الأدب: مَرَس التَّمرَ ومَرَثه: مَرَده. وفي الصحاح: الجُثْمان والجُسْمان يقال: ما أحسنَ جُثْمان الرجل وجُسْمَانه: أي جسده. وارْبَسَّ أمرهم ارْبسَاسَاً لغة في ارْبَثَّ: أي ضعف حتى تفرَّقوا. ومَرَث التمر بيده لغة في مَرَسه. وفي فقه اللغة: يقال عَثا الشيخ وعَسا. لطيفة: في الجمهرة امرأة عَثَّة بالثاء وعَشَّة بالشين المعجمة: ضئيلة الجسم، وهذا يناسب من يلثغ في الشين سيناً وفي السين ثاء وهذا يناسب: مَسَحَها بالمنديل مثل مش. والهيْثُ: الحركة مثل الهَيْشِ والهَيْثَة: الجماعة من الناس مثل الهَيْشَة.
وفي ديوان الأدب للفارابي: رجل مَغِث أي مَرِس وهذا يناسب من يلثغ في الراء والسين معا. [السيوطي، المزهر في علوم اللغة وأنواعها، النوع السابع والثلاثون، معرفة ما ورد بوجهين بحيث يؤمن فيه التصحيف]
An alternation between ð and z is seen in ðbl “wither” vs. zbl “manure, garbage,” both of which are related to bly “decay.”
في الإبدال لابن السكيت: موت ذُؤَاف وزؤاف: يعجل القتل. وزرق الطائر وذرق وزَبَرْت الكتاب وذَبَرْتُه: كَتبتُه. وفي الغريب المصنف لأبي عبيد: مر فلان وله أذْيَب وأحسبها تُقال بالزاي أيضا أزْيَب: يعني النشاط وموت ذُعاف وزُعاف مثل زؤاف. وفي ديوان الأدب: الأحْوذيّ والأحْوَزِي: الرَّاعي المشمِّر للرعاية الضابط لما وَلَى. وفي الصحاح: الأحْوَذَي مثل الأحْوزي: وهو السائق الخفيف عن أبي عمروقال العجاج: // من الرجز // (يَحُوزُهُنّ ولَهُ حُوزيُّ) وأبو عبيدة يَرويه بالذال والمعنى واحد. [السيوطي، المزهر في علوم اللغة وأنواعها، النوع السابع والثلاثون، معرفة ما ورد بوجهين بحيث يؤمن فيه التصحيف]
The initial z– in zʕl “upset, distress” must have been originally a preformative (c.f. ʕll “sick”). The same is true for zrʕ “grow plants” (c.f. rʕy “care for”) and zny “commit adultery” (c.f. nyk “have sexual intercourse”). Sometimes an initial ð– seems to have been originally a preformative, e.g. ðbħ “slaughter” vs. bħr “cut;” ðhl “astonish” vs. hwl “astonish;” ðyl “tail” vs. wly “come after;” and ðhb “go” vs. hbṭ “translocate.” The same is true for ṣ– in ṣħr “desert” vs. ħrr “hot;” ṣħb “accompany, befriend” vs. ħbb “love;” ṣħy “wake” vs. ħyy “live;” ṣrx “shout” vs. rxm “shout;” ṣbr “be patient” vs. brd “cold;” and ṣhr “melt” vs. hrs “crush.” Also š– in šbh “appear like” and šbħ “ghost” vs. bhr and bhy “shine.”
We either have to suppose that Afro-Asiatic originally had a huge array of sibilant and affricate preformatives (s–/z–/ṣ–/θ–/ð–/θ̣–/ᵗɬ–/ᵗɬ̣–), or we should suppose that all of those are just phonetic variations of a few number of original preformatives. The only real preformatives known in Afro-Asiatic grammar which could have produced all of those variations are s– and t–.
۞ SIDE NOTE
In my opinion, it is possible that the sounds *θ, *ð and *θ̣ were originally *ᵗsʸ, *ᵈzʸ and *ᵗṣʸ, that is, they were palatalized variants of the plain *ᵗs, *ᵈz and *ᵗṣ. Just like how *sʸ developed to *θ, the sounds *ᵗsʸ, *ᵈzʸ and *ᵗṣʸ developed to *ᵗθʸ , *ᵈðʸ and *ᵗθ̣ʸ. In Semitic these were deaffricated to *θʸ, *ðʸ and *θ̣ʸ, but in other Afro-Asiatic branches they were retracted to *ᵗš , *ᵈž and *ᵗṣ̌. This hypothesis reduces the six Semitic sounds θ, ð, *θ̣, *ᵗs, *ᵈz and *ᵗṣ to original three *ᵗs, *ᵈz and *ᵗṣ. It is possible that the lateral affricates were originally variants of these three as well. There is no evidence for a voiced lateral *ᵈɮ in Semitic, which is strange given that there is an emphatic lateral *ᵗɬ̣ and one would expect a voiced lateral to complete the triad. However, since the voiceless laterals *ᵗɬ and *ᵗɬ̣ alternate in some Arabic words with l, one can imagine that there was a voiced *ᵈɮ but this merged with *l in prehistoric times. This would have been analogous to the attested merger of *ᵗɬ̣ with l in some Arabic dialects of Yemen (§II.21.A.c.β.II.).
This is just speculation, but it seems that originally there were three affricates *ᵗs, *ᵈz, *ᵗṣ and a single sibilant *s. The tendency to palatalization (which is very evident in the Ancient Egyptian language) caused these four sounds to branch and produce several others.